warfarin_linUCB_PharmGKB_2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "source": [
    "import pandas as pd\n",
    "import warnings\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "import time\n",
    "\n",
    "warnings.simplefilter(action='ignore', category=FutureWarning)\n",
    "warnings.simplefilter(action='ignore', category=DeprecationWarning)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2024-05-28T09:14:55.634945Z",
     "start_time": "2024-05-28T09:14:55.632342Z"
    }
   },
   "id": "b8220aa2932ad65b",
   "outputs": [],
   "execution_count": 1
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": [
    "# Train and test on real data\n",
    "## Short range\n",
    "### Filter"
   ],
   "id": "76ff9ae1799651d5"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:14:56.333591Z",
     "start_time": "2024-05-28T09:14:55.783539Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# https://api.pharmgkb.org/v1/download/submission/553247439\n",
    "warfarin = pd.read_excel('PS206767-553247439.xls', sheet_name=1)"
   ],
   "id": "6e046944ff587c75",
   "outputs": [],
   "execution_count": 2
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:14:56.342448Z",
     "start_time": "2024-05-28T09:14:56.334921Z"
    }
   },
   "cell_type": "code",
   "source": [
    "warfarin_clean = warfarin.loc[:,\n",
    "                 [\n",
    "                     'Gender',\n",
    "                     'Race (OMB)',\n",
    "                     'Age',\n",
    "                     'Height (cm)',\n",
    "                     'Weight (kg)',\n",
    "\n",
    "                     'Indication for Warfarin Treatment',\n",
    "                     'Diabetes',\n",
    "                     'Congestive Heart Failure and/or Cardiomyopathy',\n",
    "                     'Valve Replacement',\n",
    "                     'Aspirin',\n",
    "                     'Acetaminophen or Paracetamol (Tylenol)',\n",
    "                     'Was Dose of Acetaminophen or Paracetamol (Tylenol) >1300mg/day',\n",
    "                     'Simvastatin (Zocor)',\n",
    "                     'Atorvastatin (Lipitor)',\n",
    "                     'Fluvastatin (Lescol)',\n",
    "                     'Lovastatin (Mevacor)',\n",
    "                     'Pravastatin (Pravachol)',\n",
    "                     'Rosuvastatin (Crestor)',\n",
    "                     'Cerivastatin (Baycol)',\n",
    "                     'Amiodarone (Cordarone)',\n",
    "                     'Carbamazepine (Tegretol)',\n",
    "                     'Phenytoin (Dilantin)',\n",
    "                     'Rifampin or Rifampicin',\n",
    "                     'Sulfonamide Antibiotics',\n",
    "                     'Macrolide Antibiotics',\n",
    "                     'Anti-fungal Azoles',\n",
    "                     'Herbal Medications, Vitamins, Supplements',\n",
    "                     'Current Smoker',\n",
    "\n",
    "                     'VKORC1 -1639 consensus',\n",
    "                     'VKORC1 497 consensus',\n",
    "                     'VKORC1 1173 consensus',\n",
    "                     'VKORC1 1542 consensus',\n",
    "                     'VKORC1 3730 consensus',\n",
    "                     'VKORC1 2255 consensus',\n",
    "                     'VKORC1 -4451 consensus',\n",
    "                     'CYP2C9 consensus',\n",
    "\n",
    "                     'Target INR',\n",
    "                     'Estimated Target INR Range Based on Indication',\n",
    "                     'Therapeutic Dose of Warfarin', ]]\n",
    "warfarin_clean = warfarin_clean.dropna(subset=['Height (cm)',\n",
    "                                               'Weight (kg)',\n",
    "                                               'Therapeutic Dose of Warfarin', ])"
   ],
   "id": "45f23b94f87a69c2",
   "outputs": [],
   "execution_count": 3
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Transform",
   "id": "1645f1e61b096911"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:14:56.442694Z",
     "start_time": "2024-05-28T09:14:56.343398Z"
    }
   },
   "cell_type": "code",
   "source": [
    "data = pd.DataFrame(columns=[\n",
    "    'Gender',\n",
    "    'Age in decades',\n",
    "    'Height (cm)',\n",
    "    'Weight (kg)',\n",
    "    'Race Asian',\n",
    "    'Race Black or African American',\n",
    "    'Race Caucasian or White',\n",
    "    'Race Unknown',\n",
    "\n",
    "    'DVT',\n",
    "    'PE',\n",
    "    'Afib/flutter',\n",
    "    'Heart Valve',\n",
    "    'Cardiomyopathy/LV Dilation',\n",
    "    'Stroke',\n",
    "    'Post-Orthopedic',\n",
    "    'Other',\n",
    "    'Diabetes',\n",
    "    'Congestive Heart Failure and/or Cardiomyopathy',\n",
    "    'Valve Replacement',\n",
    "\n",
    "    'Aspirin',\n",
    "    'Acetaminophen or Paracetamol (Tylenol)',\n",
    "    'Was Dose of Acetaminophen or Paracetamol (Tylenol) >1300mg/day',\n",
    "    'Simvastatin (Zocor)',\n",
    "    'Atorvastatin (Lipitor)',\n",
    "    'Fluvastatin (Lescol)',\n",
    "    'Lovastatin (Mevacor)',\n",
    "    'Pravastatin (Pravachol)',\n",
    "    'Rosuvastatin (Crestor)',\n",
    "    'Cerivastatin (Baycol)',\n",
    "    'Amiodarone (Cordarone)',\n",
    "    'Carbamazepine (Tegretol)',\n",
    "    'Phenytoin (Dilantin)',\n",
    "    'Rifampin or Rifampicin',\n",
    "    'Sulfonamide Antibiotics',\n",
    "    'Macrolide Antibiotics',\n",
    "    'Anti-fungal Azoles',\n",
    "    'Herbal Medications, Vitamins, Supplements',\n",
    "    'Current Smoker',\n",
    "\n",
    "    'CYP2C9 *1',\n",
    "    'CYP2C9 *2',\n",
    "    'CYP2C9 *3',\n",
    "    'CYP2C9 *4',\n",
    "    'CYP2C9 *5',\n",
    "    'CYP2C9 *6',\n",
    "    'CYP2C9 *7',\n",
    "    'CYP2C9 *8',\n",
    "    'CYP2C9 *9',\n",
    "    'CYP2C9 *10',\n",
    "    'CYP2C9 *11',\n",
    "    'CYP2C9 *12',\n",
    "    'CYP2C9 *13',\n",
    "    'CYP2C9 *14',\n",
    "\n",
    "    'VKORC1 - -1639G>A',\n",
    "    'VKORC1 - 497T>G',\n",
    "    'VKORC1 - 1173C>T',\n",
    "    'VKORC1 - 1542G>C',\n",
    "    'VKORC1 - 3730G>A',\n",
    "    'VKORC1 - 2255C>T',\n",
    "    'VKORC1 - -4451C>A',\n",
    "\n",
    "    'Target INR',\n",
    "    'dose',\n",
    "    'type'])"
   ],
   "id": "initial_id",
   "outputs": [],
   "execution_count": 4
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": [
    "### Type detail\n",
    "\n",
    "- Ngày 1 viên 1mg -> 7mg/tuần\n",
    "- Ngày 1 viên 2mg hoặc 2 viên 1mg -> 14mg/tuần\n",
    "- Ngày 1 viên 3mg -> 21mg/tuần\n",
    "- Ngày 1 viên 4mg hoặc 2 viên 2mg -> 28mg/tuần\n",
    "- Ngày 1 viên 5mg hoặc 2 viên 2.5mg -> 35mg/tuần\n",
    "- Ngày 1 viên 6mg hoặc 2 viên 3mg -> 42mg/tuần\n",
    "- Ngày 1 viên 7.5mg -> 52.5mg/tuần\n",
    "- Ngày 2 viên 4mg -> 56mg/tuần\n",
    "- Ngày 2 viên 5mg -> 70mg/tuần\n",
    "- Ngày 2 viên 6mg -> 84mg/tuần\n",
    "- Ngày 2 viên 7.5mg -> 105mg/tuần"
   ],
   "id": "d6b27076f57d56bd"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:14:57.531082Z",
     "start_time": "2024-05-28T09:14:56.444531Z"
    }
   },
   "cell_type": "code",
   "source": [
    "warfarin_data = []\n",
    "\n",
    "for index, row in warfarin_clean.iterrows():\n",
    "    sample = {\n",
    "        'Gender': 1 if row['Gender'] == 'male' else 0,\n",
    "        'Age in decades': 0,\n",
    "        'Height (cm)': row['Height (cm)'],\n",
    "        'Weight (kg)': row['Weight (kg)'],\n",
    "\n",
    "        'DVT': 0,\n",
    "        'PE': 0,\n",
    "        'Afib/flutter': 0,\n",
    "        'Heart Valve': 0,\n",
    "        'Cardiomyopathy/LV Dilation': 0,\n",
    "        'Stroke': 0,\n",
    "        'Post-Orthopedic': 0,\n",
    "        'Other': 0,\n",
    "        'Diabetes': 0 if pd.isnull(row['Diabetes']) else row['Diabetes'],\n",
    "        'Congestive Heart Failure and/or Cardiomyopathy': 0 if pd.isnull(row['Congestive Heart Failure and/or Cardiomyopathy']) else row['Congestive Heart Failure and/or Cardiomyopathy'],\n",
    "        'Valve Replacement': 0 if pd.isnull(row['Valve Replacement']) else row['Valve Replacement'],\n",
    "        \n",
    "        'Aspirin': 0 if pd.isnull(row['Aspirin']) else row['Aspirin'],\n",
    "        'Acetaminophen or Paracetamol (Tylenol)': 0 if pd.isnull(row['Acetaminophen or Paracetamol (Tylenol)']) else row['Acetaminophen or Paracetamol (Tylenol)'],\n",
    "        'Was Dose of Acetaminophen or Paracetamol (Tylenol) >1300mg/day': 0 if pd.isnull(row['Was Dose of Acetaminophen or Paracetamol (Tylenol) >1300mg/day']) else row['Was Dose of Acetaminophen or Paracetamol (Tylenol) >1300mg/day'],\n",
    "        'Simvastatin (Zocor)': 0 if pd.isnull(row['Simvastatin (Zocor)']) else row['Simvastatin (Zocor)'],\n",
    "        'Atorvastatin (Lipitor)': 0 if pd.isnull(row['Atorvastatin (Lipitor)']) else row['Atorvastatin (Lipitor)'],\n",
    "        'Fluvastatin (Lescol)': 0 if pd.isnull(row['Fluvastatin (Lescol)']) else row['Fluvastatin (Lescol)'],\n",
    "        'Lovastatin (Mevacor)': 0 if pd.isnull(row['Lovastatin (Mevacor)']) else row['Lovastatin (Mevacor)'],\n",
    "        'Pravastatin (Pravachol)': 0 if pd.isnull(row['Pravastatin (Pravachol)']) else row['Pravastatin (Pravachol)'],\n",
    "        'Rosuvastatin (Crestor)': 0 if pd.isnull(row['Rosuvastatin (Crestor)']) else row['Rosuvastatin (Crestor)'],\n",
    "        'Cerivastatin (Baycol)': 0 if pd.isnull(row['Cerivastatin (Baycol)']) else row['Cerivastatin (Baycol)'],\n",
    "        'Amiodarone (Cordarone)': 0 if pd.isnull(row['Amiodarone (Cordarone)']) else row['Amiodarone (Cordarone)'],\n",
    "        'Carbamazepine (Tegretol)': 0 if pd.isnull(row['Carbamazepine (Tegretol)']) else row['Carbamazepine (Tegretol)'],\n",
    "        'Phenytoin (Dilantin)': 0 if pd.isnull(row['Phenytoin (Dilantin)']) else row['Phenytoin (Dilantin)'],\n",
    "        'Rifampin or Rifampicin': 0 if pd.isnull(row['Rifampin or Rifampicin']) else row['Rifampin or Rifampicin'],\n",
    "        'Sulfonamide Antibiotics': 0 if pd.isnull(row['Sulfonamide Antibiotics']) else row['Sulfonamide Antibiotics'],\n",
    "        'Macrolide Antibiotics': 0 if pd.isnull(row['Macrolide Antibiotics']) else row['Macrolide Antibiotics'],\n",
    "        'Anti-fungal Azoles': 0 if pd.isnull(row['Anti-fungal Azoles']) else row['Anti-fungal Azoles'],\n",
    "        'Herbal Medications, Vitamins, Supplements': 0 if pd.isnull(row['Herbal Medications, Vitamins, Supplements']) else row['Herbal Medications, Vitamins, Supplements'],\n",
    "        'Current Smoker': 0 if pd.isnull(row['Current Smoker']) else row['Current Smoker'],\n",
    "    \n",
    "        'Race Asian': 0,\n",
    "        'Race Black or African American': 0,\n",
    "        'Race Unknown': 0,\n",
    "        'Race Caucasian or White': 0,\n",
    "\n",
    "        'CYP2C9 *1': 0,\n",
    "        'CYP2C9 *2': 0,\n",
    "        'CYP2C9 *3': 0,\n",
    "        'CYP2C9 *4': 0,\n",
    "        'CYP2C9 *5': 0,\n",
    "        'CYP2C9 *6': 0,\n",
    "        'CYP2C9 *7': 0,\n",
    "        'CYP2C9 *8': 0,\n",
    "        'CYP2C9 *9': 0,\n",
    "        'CYP2C9 *10': 0,\n",
    "        'CYP2C9 *11': 0,\n",
    "        'CYP2C9 *12': 0,\n",
    "        'CYP2C9 *13': 0,\n",
    "        'CYP2C9 *14': 0,\n",
    "    \n",
    "        'VKORC1 - -1639G>A': 0,\n",
    "        'VKORC1 - 497T>G': 0,\n",
    "        'VKORC1 - 1173C>T': 0,\n",
    "        'VKORC1 - 1542G>C': 0,\n",
    "        'VKORC1 - 3730G>A': 0,\n",
    "        'VKORC1 - 2255C>T': 0,\n",
    "        'VKORC1 - -4451C>A': 0,\n",
    "\n",
    "        'Target INR': 0,\n",
    "        'dose': row['Therapeutic Dose of Warfarin'],\n",
    "        'type': 0\n",
    "    }\n",
    "    try:\n",
    "        sample['Age in decades'] = int(str(row['Age']).split('-')[0].replace('+', '')) / 10\n",
    "    except:\n",
    "        continue\n",
    "\n",
    "    if pd.isnull(row['Indication for Warfarin Treatment']):\n",
    "        sample['Other']=1\n",
    "    else:\n",
    "        sample['DVT'] = str(row['Indication for Warfarin Treatment']).count('1')\n",
    "        sample['PE'] = str(row['Indication for Warfarin Treatment']).count('2')\n",
    "        sample['Afib/flutter'] = str(row['Indication for Warfarin Treatment']).count('3')\n",
    "        sample['Heart Valve'] = str(row['Indication for Warfarin Treatment']).count('4')\n",
    "        sample['Cardiomyopathy/LV Dilation'] = str(row['Indication for Warfarin Treatment']).count('5')\n",
    "        sample['Stroke'] = str(row['Indication for Warfarin Treatment']).count('6')\n",
    "        sample['Post-Orthopedic'] = str(row['Indication for Warfarin Treatment']).count('7')\n",
    "        sample['Other'] = str(row['Indication for Warfarin Treatment']).count('8')\n",
    "\n",
    "    match row['Race (OMB)']:\n",
    "        case 'White':\n",
    "            sample['Race Caucasian or White'] = 1\n",
    "        case 'Asian':\n",
    "            sample['Race Asian'] = 1\n",
    "        case 'Black or African American':\n",
    "            sample['Race Black or African American'] = 1\n",
    "        case 'Unknown':\n",
    "            sample['Race Unknown'] = 1\n",
    "\n",
    "    if pd.isnull(row['VKORC1 -1639 consensus']):\n",
    "        pass\n",
    "    else:\n",
    "        sample['VKORC1 - -1639G>A'] = str(row['VKORC1 -1639 consensus']).count('A')\n",
    "        \n",
    "    if pd.isnull(row['VKORC1 497 consensus']):\n",
    "        pass\n",
    "    else:\n",
    "        sample['VKORC1 - 497T>G'] = str(row['VKORC1 497 consensus']).count('G')\n",
    "        \n",
    "    if pd.isnull(row['VKORC1 1173 consensus']):\n",
    "        pass\n",
    "    else:\n",
    "        sample['VKORC1 - 1173C>T'] = str(row['VKORC1 1173 consensus']).count('T')\n",
    "        \n",
    "    if pd.isnull(row['VKORC1 1542 consensus']):\n",
    "        pass\n",
    "    else:\n",
    "        sample['VKORC1 - 1542G>C'] = str(row['VKORC1 1542 consensus']).count('C')\n",
    "        \n",
    "    if pd.isnull(row['VKORC1 3730 consensus']):\n",
    "        pass\n",
    "    else:\n",
    "        sample['VKORC1 - 3730G>A'] = str(row['VKORC1 3730 consensus']).count('A')\n",
    "        \n",
    "    if pd.isnull(row['VKORC1 2255 consensus']):\n",
    "        pass\n",
    "    else:\n",
    "        sample['VKORC1 - 2255C>T'] = str(row['VKORC1 2255 consensus']).count('T')\n",
    "        \n",
    "    if pd.isnull(row['VKORC1 -4451 consensus']):\n",
    "        pass\n",
    "    else:\n",
    "        sample['VKORC1 - -4451C>A'] = str(row['VKORC1 -4451 consensus']).count('A')\n",
    "\n",
    "    sample['CYP2C9 *1'] = str(row['CYP2C9 consensus']).count('*1')\n",
    "    sample['CYP2C9 *2'] = str(row['CYP2C9 consensus']).count('*2')\n",
    "    sample['CYP2C9 *3'] = str(row['CYP2C9 consensus']).count('*3')\n",
    "    sample['CYP2C9 *4'] = str(row['CYP2C9 consensus']).count('*4')\n",
    "    sample['CYP2C9 *5'] = str(row['CYP2C9 consensus']).count('*5')\n",
    "    sample['CYP2C9 *6'] = str(row['CYP2C9 consensus']).count('*6')\n",
    "    sample['CYP2C9 *7'] = str(row['CYP2C9 consensus']).count('*7')\n",
    "    sample['CYP2C9 *8'] = str(row['CYP2C9 consensus']).count('*8')\n",
    "    sample['CYP2C9 *9'] = str(row['CYP2C9 consensus']).count('*9')\n",
    "    sample['CYP2C9 *10'] = str(row['CYP2C9 consensus']).count('*10')\n",
    "    sample['CYP2C9 *11'] = str(row['CYP2C9 consensus']).count('*11')\n",
    "    sample['CYP2C9 *12'] = str(row['CYP2C9 consensus']).count('*12')\n",
    "    sample['CYP2C9 *13'] = str(row['CYP2C9 consensus']).count('*13')\n",
    "    sample['CYP2C9 *14'] = str(row['CYP2C9 consensus']).count('*14')\n",
    "\n",
    "    if not pd.isnull(row['Target INR']):\n",
    "        sample['Target INR'] = row['Target INR']\n",
    "    else:\n",
    "        if not pd.isnull(row['Estimated Target INR Range Based on Indication']):\n",
    "            sample['Target INR'] = float(str(row['Estimated Target INR Range Based on Indication']).split('-')[1])\n",
    "        else:\n",
    "            continue\n",
    "    \n",
    "    if sample['dose'] < 7:\n",
    "        sample['type'] = 0\n",
    "    elif 7 <= sample['dose'] < 14:\n",
    "        sample['type'] = 1\n",
    "    elif 14 <= sample['dose'] < 21:\n",
    "        sample['type'] = 2\n",
    "    elif 21 <= sample['dose'] < 28:\n",
    "        sample['type'] = 3\n",
    "    elif 28 <= sample['dose'] < 35:\n",
    "        sample['type'] = 4\n",
    "    elif 35 <= sample['dose'] < 42:\n",
    "        sample['type'] = 5\n",
    "    elif 42 <= sample['dose'] < 52.5:\n",
    "        sample['type'] = 6\n",
    "    elif 52.5 <= sample['dose'] < 56:\n",
    "        sample['type'] = 7\n",
    "    elif 56 <= sample['dose'] < 70:\n",
    "        sample['type'] = 8\n",
    "    elif 70 <= sample['dose'] < 84:\n",
    "        sample['type'] = 9\n",
    "    elif 84 <= sample['dose'] < 105:\n",
    "        sample['type'] = 10\n",
    "    else:\n",
    "        sample['type'] = 11\n",
    "    warfarin_data.append(sample)\n",
    "\n",
    "data = data._append(warfarin_data, ignore_index=True)"
   ],
   "id": "8bb79d9c1ce668ad",
   "outputs": [],
   "execution_count": 5
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:14:57.541743Z",
     "start_time": "2024-05-28T09:14:57.535770Z"
    }
   },
   "cell_type": "code",
   "source": "data.shape",
   "id": "7129b75a92466bcc",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3663, 62)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "execution_count": 6
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:14:57.637822Z",
     "start_time": "2024-05-28T09:14:57.542817Z"
    }
   },
   "cell_type": "code",
   "source": "data.info()",
   "id": "f5ffe7b5d53c7efa",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 3663 entries, 0 to 3662\n",
      "Data columns (total 62 columns):\n",
      " #   Column                                                          Non-Null Count  Dtype  \n",
      "---  ------                                                          --------------  -----  \n",
      " 0   Gender                                                          3663 non-null   object \n",
      " 1   Age in decades                                                  3663 non-null   float64\n",
      " 2   Height (cm)                                                     3663 non-null   float64\n",
      " 3   Weight (kg)                                                     3663 non-null   float64\n",
      " 4   Race Asian                                                      3663 non-null   object \n",
      " 5   Race Black or African American                                  3663 non-null   object \n",
      " 6   Race Caucasian or White                                         3663 non-null   object \n",
      " 7   Race Unknown                                                    3663 non-null   object \n",
      " 8   DVT                                                             3663 non-null   object \n",
      " 9   PE                                                              3663 non-null   object \n",
      " 10  Afib/flutter                                                    3663 non-null   object \n",
      " 11  Heart Valve                                                     3663 non-null   object \n",
      " 12  Cardiomyopathy/LV Dilation                                      3663 non-null   object \n",
      " 13  Stroke                                                          3663 non-null   object \n",
      " 14  Post-Orthopedic                                                 3663 non-null   object \n",
      " 15  Other                                                           3663 non-null   object \n",
      " 16  Diabetes                                                        3663 non-null   float64\n",
      " 17  Congestive Heart Failure and/or Cardiomyopathy                  3663 non-null   float64\n",
      " 18  Valve Replacement                                               3663 non-null   float64\n",
      " 19  Aspirin                                                         3663 non-null   float64\n",
      " 20  Acetaminophen or Paracetamol (Tylenol)                          3663 non-null   float64\n",
      " 21  Was Dose of Acetaminophen or Paracetamol (Tylenol) >1300mg/day  3663 non-null   float64\n",
      " 22  Simvastatin (Zocor)                                             3663 non-null   float64\n",
      " 23  Atorvastatin (Lipitor)                                          3663 non-null   float64\n",
      " 24  Fluvastatin (Lescol)                                            3663 non-null   float64\n",
      " 25  Lovastatin (Mevacor)                                            3663 non-null   float64\n",
      " 26  Pravastatin (Pravachol)                                         3663 non-null   float64\n",
      " 27  Rosuvastatin (Crestor)                                          3663 non-null   float64\n",
      " 28  Cerivastatin (Baycol)                                           3663 non-null   float64\n",
      " 29  Amiodarone (Cordarone)                                          3663 non-null   float64\n",
      " 30  Carbamazepine (Tegretol)                                        3663 non-null   float64\n",
      " 31  Phenytoin (Dilantin)                                            3663 non-null   float64\n",
      " 32  Rifampin or Rifampicin                                          3663 non-null   float64\n",
      " 33  Sulfonamide Antibiotics                                         3663 non-null   float64\n",
      " 34  Macrolide Antibiotics                                           3663 non-null   float64\n",
      " 35  Anti-fungal Azoles                                              3663 non-null   float64\n",
      " 36  Herbal Medications, Vitamins, Supplements                       3663 non-null   float64\n",
      " 37  Current Smoker                                                  3663 non-null   float64\n",
      " 38  CYP2C9 *1                                                       3663 non-null   object \n",
      " 39  CYP2C9 *2                                                       3663 non-null   object \n",
      " 40  CYP2C9 *3                                                       3663 non-null   object \n",
      " 41  CYP2C9 *4                                                       3663 non-null   object \n",
      " 42  CYP2C9 *5                                                       3663 non-null   object \n",
      " 43  CYP2C9 *6                                                       3663 non-null   object \n",
      " 44  CYP2C9 *7                                                       3663 non-null   object \n",
      " 45  CYP2C9 *8                                                       3663 non-null   object \n",
      " 46  CYP2C9 *9                                                       3663 non-null   object \n",
      " 47  CYP2C9 *10                                                      3663 non-null   object \n",
      " 48  CYP2C9 *11                                                      3663 non-null   object \n",
      " 49  CYP2C9 *12                                                      3663 non-null   object \n",
      " 50  CYP2C9 *13                                                      3663 non-null   object \n",
      " 51  CYP2C9 *14                                                      3663 non-null   object \n",
      " 52  VKORC1 - -1639G>A                                               3663 non-null   object \n",
      " 53  VKORC1 - 497T>G                                                 3663 non-null   object \n",
      " 54  VKORC1 - 1173C>T                                                3663 non-null   object \n",
      " 55  VKORC1 - 1542G>C                                                3663 non-null   object \n",
      " 56  VKORC1 - 3730G>A                                                3663 non-null   object \n",
      " 57  VKORC1 - 2255C>T                                                3663 non-null   object \n",
      " 58  VKORC1 - -4451C>A                                               3663 non-null   object \n",
      " 59  Target INR                                                      3663 non-null   float64\n",
      " 60  dose                                                            3663 non-null   float64\n",
      " 61  type                                                            3663 non-null   object \n",
      "dtypes: float64(27), object(35)\n",
      "memory usage: 1.7+ MB\n"
     ]
    }
   ],
   "execution_count": 7
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### split",
   "id": "9fe339f28967ad2e"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:14:57.715426Z",
     "start_time": "2024-05-28T09:14:57.639113Z"
    }
   },
   "cell_type": "code",
   "source": [
    "data_size = len(data)\n",
    "ratio = 0.7\n",
    "pointer = int(ratio * data_size)\n",
    "train_set = data.iloc[:pointer, :-2]\n",
    "test_set = data.iloc[pointer:, :-2]\n",
    "train_dose = data.iloc[:pointer, -1:].values\n",
    "test_dose = data.iloc[pointer:, -1:].values"
   ],
   "id": "36c424f1385a05ec",
   "outputs": [],
   "execution_count": 8
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### train",
   "id": "c6bfc77c160b4971"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:38:46.740964Z",
     "start_time": "2024-05-28T09:14:57.716696Z"
    }
   },
   "cell_type": "code",
   "source": [
    "arms = np.arange(12)\n",
    "accuracy = []\n",
    "A_map = {}\n",
    "b_map = {}\n",
    "for arm in arms:\n",
    "    A_map[arm] = np.identity(len(list(train_set)))\n",
    "    b_map[arm] = np.zeros(len(list(train_set)))\n",
    "\n",
    "for epoch in range(200):\n",
    "    start_time = time.time()\n",
    "    # train\n",
    "    indices = np.random.permutation(list(range(len(train_set))))\n",
    "    for i in indices:\n",
    "        feature = np.array(train_set.iloc[i], dtype=float)\n",
    "        # iterate through actions\n",
    "        p = {}\n",
    "        max_value = -np.inf\n",
    "        max_arm = 0\n",
    "        for arm in arms:\n",
    "            theta = np.matmul(np.linalg.inv(A_map[arm]), b_map[arm])\n",
    "            p = np.matmul(theta.T, feature) + \\\n",
    "                np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map[arm])), feature))\n",
    "\n",
    "            if p == max_value:\n",
    "                if np.random.choice(a=[False, True]): max_arm = arm\n",
    "            elif p > max_value:\n",
    "                max_value = p\n",
    "                max_arm = arm\n",
    "\n",
    "        # observe reward and update A and b\n",
    "        r = -abs(train_dose[i][0]-max_arm)\n",
    "        A_map[max_arm] += np.outer(feature, feature)\n",
    "        b_map[max_arm] += r * feature\n",
    "\n",
    "    # test\n",
    "    correct_count = 0\n",
    "    count = 0\n",
    "    higher = 0\n",
    "    lower = 0\n",
    "    for i in range(len(test_set)):\n",
    "        count += 1\n",
    "        feature = np.array(test_set.iloc[i], dtype=float)\n",
    "        # iterate through actions\n",
    "        p = {}\n",
    "        max_value = -np.inf\n",
    "        max_arm = 0\n",
    "        for arm in arms:\n",
    "            theta = np.matmul(np.linalg.inv(A_map[arm]), b_map[arm])\n",
    "            p = np.matmul(theta.T, feature) + \\\n",
    "                np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map[arm])), feature))\n",
    "            if p == max_value:\n",
    "                if np.random.choice(a=[False, True]): max_arm = arm\n",
    "            elif p > max_value:\n",
    "                max_value = p\n",
    "                max_arm = arm\n",
    "\n",
    "        if max_arm == test_dose[i]:\n",
    "            correct_count += 1\n",
    "        elif max_arm < test_dose[i]:\n",
    "            lower += test_dose[i] - max_arm\n",
    "        else:\n",
    "            higher += max_arm - test_dose[i]\n",
    "\n",
    "    end_time = time.time()\n",
    "    accu = correct_count / count\n",
    "    print(\"Epoch %d: Accuracy: %.5f%% (%d/%d), Higher: %d, Lower: %d, Time: %.5fs\" % (\n",
    "        epoch, accu * 100, correct_count, count, higher, lower, end_time - start_time))\n",
    "    accuracy.append(accu)"
   ],
   "id": "ef0157a1766ae6da",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0: Accuracy: 20.83712% (229/1099), Higher: 1248, Lower: 764, Time: 7.67767s\n",
      "Epoch 1: Accuracy: 21.29208% (234/1099), Higher: 1016, Lower: 781, Time: 7.41043s\n",
      "Epoch 2: Accuracy: 20.92812% (230/1099), Higher: 1061, Lower: 756, Time: 6.99112s\n",
      "Epoch 3: Accuracy: 20.83712% (229/1099), Higher: 1084, Lower: 751, Time: 7.08107s\n",
      "Epoch 4: Accuracy: 21.65605% (238/1099), Higher: 1058, Lower: 773, Time: 6.90363s\n",
      "Epoch 5: Accuracy: 21.56506% (237/1099), Higher: 1055, Lower: 765, Time: 6.86828s\n",
      "Epoch 6: Accuracy: 21.01911% (231/1099), Higher: 1084, Lower: 788, Time: 7.01315s\n",
      "Epoch 7: Accuracy: 20.38217% (224/1099), Higher: 1094, Lower: 788, Time: 7.13308s\n",
      "Epoch 8: Accuracy: 20.29117% (223/1099), Higher: 1078, Lower: 807, Time: 7.61910s\n",
      "Epoch 9: Accuracy: 21.01911% (231/1099), Higher: 1073, Lower: 800, Time: 7.52573s\n",
      "Epoch 10: Accuracy: 20.92812% (230/1099), Higher: 1085, Lower: 806, Time: 7.56503s\n",
      "Epoch 11: Accuracy: 21.29208% (234/1099), Higher: 1073, Lower: 808, Time: 7.13451s\n",
      "Epoch 12: Accuracy: 21.56506% (237/1099), Higher: 1081, Lower: 793, Time: 6.90120s\n",
      "Epoch 13: Accuracy: 21.65605% (238/1099), Higher: 1072, Lower: 800, Time: 6.96940s\n",
      "Epoch 14: Accuracy: 21.56506% (237/1099), Higher: 1072, Lower: 798, Time: 7.07200s\n",
      "Epoch 15: Accuracy: 21.29208% (234/1099), Higher: 1070, Lower: 808, Time: 7.16599s\n",
      "Epoch 16: Accuracy: 21.20109% (233/1099), Higher: 1067, Lower: 809, Time: 7.11288s\n",
      "Epoch 17: Accuracy: 21.01911% (231/1099), Higher: 1043, Lower: 806, Time: 7.02331s\n",
      "Epoch 18: Accuracy: 20.83712% (229/1099), Higher: 1061, Lower: 806, Time: 7.10234s\n",
      "Epoch 19: Accuracy: 21.01911% (231/1099), Higher: 1059, Lower: 800, Time: 7.06154s\n",
      "Epoch 20: Accuracy: 20.83712% (229/1099), Higher: 1066, Lower: 801, Time: 7.15352s\n",
      "Epoch 21: Accuracy: 21.11010% (232/1099), Higher: 1053, Lower: 803, Time: 7.01246s\n",
      "Epoch 22: Accuracy: 20.83712% (229/1099), Higher: 1055, Lower: 807, Time: 7.13776s\n",
      "Epoch 23: Accuracy: 20.74613% (228/1099), Higher: 1058, Lower: 812, Time: 7.02997s\n",
      "Epoch 24: Accuracy: 20.56415% (226/1099), Higher: 1059, Lower: 812, Time: 7.05037s\n",
      "Epoch 25: Accuracy: 20.47316% (225/1099), Higher: 1047, Lower: 816, Time: 7.15967s\n",
      "Epoch 26: Accuracy: 20.38217% (224/1099), Higher: 1049, Lower: 818, Time: 7.10285s\n",
      "Epoch 27: Accuracy: 20.20018% (222/1099), Higher: 1042, Lower: 823, Time: 6.93143s\n",
      "Epoch 28: Accuracy: 20.38217% (224/1099), Higher: 1040, Lower: 817, Time: 7.06000s\n",
      "Epoch 29: Accuracy: 20.01820% (220/1099), Higher: 1039, Lower: 823, Time: 7.03749s\n",
      "Epoch 30: Accuracy: 20.20018% (222/1099), Higher: 1040, Lower: 823, Time: 7.11158s\n",
      "Epoch 31: Accuracy: 20.20018% (222/1099), Higher: 1039, Lower: 822, Time: 7.03446s\n",
      "Epoch 32: Accuracy: 20.29117% (223/1099), Higher: 1049, Lower: 821, Time: 6.99959s\n",
      "Epoch 33: Accuracy: 20.10919% (221/1099), Higher: 1049, Lower: 825, Time: 6.97224s\n",
      "Epoch 34: Accuracy: 20.20018% (222/1099), Higher: 1047, Lower: 826, Time: 7.01968s\n",
      "Epoch 35: Accuracy: 20.20018% (222/1099), Higher: 1040, Lower: 828, Time: 7.00058s\n",
      "Epoch 36: Accuracy: 20.10919% (221/1099), Higher: 1041, Lower: 831, Time: 7.14559s\n",
      "Epoch 37: Accuracy: 20.47316% (225/1099), Higher: 1024, Lower: 829, Time: 6.97355s\n",
      "Epoch 38: Accuracy: 20.38217% (224/1099), Higher: 1036, Lower: 828, Time: 6.94342s\n",
      "Epoch 39: Accuracy: 20.29117% (223/1099), Higher: 1037, Lower: 826, Time: 6.87629s\n",
      "Epoch 40: Accuracy: 20.29117% (223/1099), Higher: 1034, Lower: 828, Time: 6.88824s\n",
      "Epoch 41: Accuracy: 20.38217% (224/1099), Higher: 1029, Lower: 830, Time: 7.02675s\n",
      "Epoch 42: Accuracy: 20.38217% (224/1099), Higher: 1029, Lower: 829, Time: 7.02235s\n",
      "Epoch 43: Accuracy: 20.38217% (224/1099), Higher: 1027, Lower: 837, Time: 7.13137s\n",
      "Epoch 44: Accuracy: 20.38217% (224/1099), Higher: 1029, Lower: 831, Time: 7.20808s\n",
      "Epoch 45: Accuracy: 20.47316% (225/1099), Higher: 1029, Lower: 824, Time: 7.16609s\n",
      "Epoch 46: Accuracy: 20.47316% (225/1099), Higher: 1029, Lower: 825, Time: 7.12125s\n",
      "Epoch 47: Accuracy: 20.47316% (225/1099), Higher: 1033, Lower: 829, Time: 7.01498s\n",
      "Epoch 48: Accuracy: 20.47316% (225/1099), Higher: 1032, Lower: 828, Time: 6.95440s\n",
      "Epoch 49: Accuracy: 20.47316% (225/1099), Higher: 1032, Lower: 831, Time: 7.02433s\n",
      "Epoch 50: Accuracy: 20.38217% (224/1099), Higher: 1031, Lower: 830, Time: 6.96538s\n",
      "Epoch 51: Accuracy: 20.38217% (224/1099), Higher: 1038, Lower: 827, Time: 7.01358s\n",
      "Epoch 52: Accuracy: 20.38217% (224/1099), Higher: 1028, Lower: 836, Time: 6.93407s\n",
      "Epoch 53: Accuracy: 20.38217% (224/1099), Higher: 1026, Lower: 836, Time: 7.13312s\n",
      "Epoch 54: Accuracy: 20.47316% (225/1099), Higher: 1025, Lower: 836, Time: 6.95241s\n",
      "Epoch 55: Accuracy: 20.29117% (223/1099), Higher: 1028, Lower: 839, Time: 7.06978s\n",
      "Epoch 56: Accuracy: 20.29117% (223/1099), Higher: 1024, Lower: 841, Time: 6.92507s\n",
      "Epoch 57: Accuracy: 20.29117% (223/1099), Higher: 1028, Lower: 841, Time: 7.07074s\n",
      "Epoch 58: Accuracy: 20.20018% (222/1099), Higher: 1027, Lower: 840, Time: 7.04826s\n",
      "Epoch 59: Accuracy: 20.29117% (223/1099), Higher: 1023, Lower: 838, Time: 7.04464s\n",
      "Epoch 60: Accuracy: 20.20018% (222/1099), Higher: 1025, Lower: 841, Time: 6.98068s\n",
      "Epoch 61: Accuracy: 20.10919% (221/1099), Higher: 1019, Lower: 845, Time: 7.04406s\n",
      "Epoch 62: Accuracy: 20.20018% (222/1099), Higher: 1018, Lower: 842, Time: 7.08043s\n",
      "Epoch 63: Accuracy: 20.20018% (222/1099), Higher: 1023, Lower: 847, Time: 7.01370s\n",
      "Epoch 64: Accuracy: 20.20018% (222/1099), Higher: 1011, Lower: 851, Time: 7.09959s\n",
      "Epoch 65: Accuracy: 20.38217% (224/1099), Higher: 1015, Lower: 849, Time: 6.95259s\n",
      "Epoch 66: Accuracy: 20.38217% (224/1099), Higher: 1013, Lower: 849, Time: 6.97677s\n",
      "Epoch 67: Accuracy: 20.38217% (224/1099), Higher: 1012, Lower: 850, Time: 6.93807s\n",
      "Epoch 68: Accuracy: 20.38217% (224/1099), Higher: 1012, Lower: 857, Time: 6.94339s\n",
      "Epoch 69: Accuracy: 20.20018% (222/1099), Higher: 1015, Lower: 853, Time: 7.25764s\n",
      "Epoch 70: Accuracy: 20.29117% (223/1099), Higher: 1013, Lower: 854, Time: 7.23914s\n",
      "Epoch 71: Accuracy: 20.29117% (223/1099), Higher: 1011, Lower: 861, Time: 7.09412s\n",
      "Epoch 72: Accuracy: 20.38217% (224/1099), Higher: 1007, Lower: 862, Time: 7.19497s\n",
      "Epoch 73: Accuracy: 20.20018% (222/1099), Higher: 1010, Lower: 862, Time: 6.97580s\n",
      "Epoch 74: Accuracy: 20.10919% (221/1099), Higher: 1008, Lower: 865, Time: 7.05023s\n",
      "Epoch 75: Accuracy: 20.10919% (221/1099), Higher: 1007, Lower: 866, Time: 7.06319s\n",
      "Epoch 76: Accuracy: 20.10919% (221/1099), Higher: 1008, Lower: 866, Time: 7.09131s\n",
      "Epoch 77: Accuracy: 20.10919% (221/1099), Higher: 1005, Lower: 866, Time: 7.17757s\n",
      "Epoch 78: Accuracy: 20.10919% (221/1099), Higher: 1004, Lower: 866, Time: 7.12093s\n",
      "Epoch 79: Accuracy: 20.20018% (222/1099), Higher: 1004, Lower: 865, Time: 7.13383s\n",
      "Epoch 80: Accuracy: 20.20018% (222/1099), Higher: 1004, Lower: 871, Time: 7.12020s\n",
      "Epoch 81: Accuracy: 20.20018% (222/1099), Higher: 1006, Lower: 870, Time: 7.08031s\n",
      "Epoch 82: Accuracy: 20.29117% (223/1099), Higher: 1005, Lower: 870, Time: 7.14317s\n",
      "Epoch 83: Accuracy: 20.20018% (222/1099), Higher: 1006, Lower: 871, Time: 7.02417s\n",
      "Epoch 84: Accuracy: 20.29117% (223/1099), Higher: 1006, Lower: 870, Time: 7.14924s\n",
      "Epoch 85: Accuracy: 20.29117% (223/1099), Higher: 1005, Lower: 870, Time: 7.02310s\n",
      "Epoch 86: Accuracy: 20.29117% (223/1099), Higher: 1006, Lower: 870, Time: 7.11758s\n",
      "Epoch 87: Accuracy: 20.29117% (223/1099), Higher: 1005, Lower: 870, Time: 7.19078s\n",
      "Epoch 88: Accuracy: 20.29117% (223/1099), Higher: 1008, Lower: 868, Time: 7.18903s\n",
      "Epoch 89: Accuracy: 20.29117% (223/1099), Higher: 1008, Lower: 868, Time: 7.00410s\n",
      "Epoch 90: Accuracy: 20.29117% (223/1099), Higher: 1008, Lower: 868, Time: 7.18634s\n",
      "Epoch 91: Accuracy: 20.29117% (223/1099), Higher: 1009, Lower: 864, Time: 7.29422s\n",
      "Epoch 92: Accuracy: 20.20018% (222/1099), Higher: 1010, Lower: 864, Time: 7.37201s\n",
      "Epoch 93: Accuracy: 20.29117% (223/1099), Higher: 1012, Lower: 864, Time: 7.09388s\n",
      "Epoch 94: Accuracy: 20.29117% (223/1099), Higher: 1012, Lower: 864, Time: 7.10937s\n",
      "Epoch 95: Accuracy: 20.20018% (222/1099), Higher: 1013, Lower: 868, Time: 7.25071s\n",
      "Epoch 96: Accuracy: 20.29117% (223/1099), Higher: 1012, Lower: 864, Time: 7.07285s\n",
      "Epoch 97: Accuracy: 20.29117% (223/1099), Higher: 1012, Lower: 868, Time: 7.03942s\n",
      "Epoch 98: Accuracy: 20.20018% (222/1099), Higher: 1013, Lower: 868, Time: 7.05922s\n",
      "Epoch 99: Accuracy: 20.20018% (222/1099), Higher: 1013, Lower: 870, Time: 6.95887s\n",
      "Epoch 100: Accuracy: 20.29117% (223/1099), Higher: 1012, Lower: 870, Time: 7.04539s\n",
      "Epoch 101: Accuracy: 20.38217% (224/1099), Higher: 1010, Lower: 870, Time: 7.00533s\n",
      "Epoch 102: Accuracy: 20.38217% (224/1099), Higher: 1010, Lower: 870, Time: 7.02283s\n",
      "Epoch 103: Accuracy: 20.29117% (223/1099), Higher: 1023, Lower: 867, Time: 7.05993s\n",
      "Epoch 104: Accuracy: 20.29117% (223/1099), Higher: 1005, Lower: 872, Time: 7.20070s\n",
      "Epoch 105: Accuracy: 20.38217% (224/1099), Higher: 1004, Lower: 871, Time: 7.03141s\n",
      "Epoch 106: Accuracy: 20.29117% (223/1099), Higher: 1005, Lower: 867, Time: 6.99909s\n",
      "Epoch 107: Accuracy: 20.29117% (223/1099), Higher: 1012, Lower: 870, Time: 7.03654s\n",
      "Epoch 108: Accuracy: 20.20018% (222/1099), Higher: 1012, Lower: 871, Time: 7.18676s\n",
      "Epoch 109: Accuracy: 20.38217% (224/1099), Higher: 1006, Lower: 871, Time: 7.09206s\n",
      "Epoch 110: Accuracy: 20.38217% (224/1099), Higher: 1006, Lower: 871, Time: 7.05304s\n",
      "Epoch 111: Accuracy: 20.38217% (224/1099), Higher: 1006, Lower: 871, Time: 6.92434s\n",
      "Epoch 112: Accuracy: 20.29117% (223/1099), Higher: 1006, Lower: 874, Time: 6.97914s\n",
      "Epoch 113: Accuracy: 20.29117% (223/1099), Higher: 1011, Lower: 873, Time: 7.00375s\n",
      "Epoch 114: Accuracy: 20.29117% (223/1099), Higher: 1011, Lower: 873, Time: 6.95056s\n",
      "Epoch 115: Accuracy: 20.38217% (224/1099), Higher: 1011, Lower: 871, Time: 7.21283s\n",
      "Epoch 116: Accuracy: 20.29117% (223/1099), Higher: 1011, Lower: 873, Time: 7.12251s\n",
      "Epoch 117: Accuracy: 20.29117% (223/1099), Higher: 1011, Lower: 871, Time: 7.05096s\n",
      "Epoch 118: Accuracy: 20.47316% (225/1099), Higher: 1002, Lower: 873, Time: 7.04697s\n",
      "Epoch 119: Accuracy: 20.38217% (224/1099), Higher: 1006, Lower: 873, Time: 7.14268s\n",
      "Epoch 120: Accuracy: 20.47316% (225/1099), Higher: 1002, Lower: 873, Time: 7.12892s\n",
      "Epoch 121: Accuracy: 20.38217% (224/1099), Higher: 1006, Lower: 873, Time: 7.26214s\n",
      "Epoch 122: Accuracy: 20.47316% (225/1099), Higher: 1002, Lower: 873, Time: 7.03254s\n",
      "Epoch 123: Accuracy: 20.47316% (225/1099), Higher: 1000, Lower: 874, Time: 7.07501s\n",
      "Epoch 124: Accuracy: 20.65514% (227/1099), Higher: 994, Lower: 874, Time: 7.12958s\n",
      "Epoch 125: Accuracy: 20.65514% (227/1099), Higher: 992, Lower: 874, Time: 7.09506s\n",
      "Epoch 126: Accuracy: 20.56415% (226/1099), Higher: 996, Lower: 874, Time: 6.99361s\n",
      "Epoch 127: Accuracy: 20.56415% (226/1099), Higher: 996, Lower: 870, Time: 7.08505s\n",
      "Epoch 128: Accuracy: 20.65514% (227/1099), Higher: 993, Lower: 870, Time: 6.97218s\n",
      "Epoch 129: Accuracy: 20.65514% (227/1099), Higher: 995, Lower: 868, Time: 7.10660s\n",
      "Epoch 130: Accuracy: 20.56415% (226/1099), Higher: 994, Lower: 869, Time: 6.99381s\n",
      "Epoch 131: Accuracy: 20.56415% (226/1099), Higher: 993, Lower: 869, Time: 7.06709s\n",
      "Epoch 132: Accuracy: 20.38217% (224/1099), Higher: 996, Lower: 874, Time: 7.07157s\n",
      "Epoch 133: Accuracy: 20.47316% (225/1099), Higher: 996, Lower: 869, Time: 6.98251s\n",
      "Epoch 134: Accuracy: 20.47316% (225/1099), Higher: 995, Lower: 869, Time: 7.06983s\n",
      "Epoch 135: Accuracy: 20.38217% (224/1099), Higher: 994, Lower: 870, Time: 6.99854s\n",
      "Epoch 136: Accuracy: 20.29117% (223/1099), Higher: 995, Lower: 871, Time: 7.16506s\n",
      "Epoch 137: Accuracy: 20.38217% (224/1099), Higher: 995, Lower: 870, Time: 7.18317s\n",
      "Epoch 138: Accuracy: 20.38217% (224/1099), Higher: 995, Lower: 870, Time: 6.96673s\n",
      "Epoch 139: Accuracy: 20.38217% (224/1099), Higher: 995, Lower: 869, Time: 7.20147s\n",
      "Epoch 140: Accuracy: 20.47316% (225/1099), Higher: 995, Lower: 868, Time: 7.18623s\n",
      "Epoch 141: Accuracy: 20.38217% (224/1099), Higher: 992, Lower: 868, Time: 7.32882s\n",
      "Epoch 142: Accuracy: 20.29117% (223/1099), Higher: 996, Lower: 869, Time: 7.25696s\n",
      "Epoch 143: Accuracy: 20.38217% (224/1099), Higher: 996, Lower: 867, Time: 7.76979s\n",
      "Epoch 144: Accuracy: 20.38217% (224/1099), Higher: 992, Lower: 868, Time: 7.18204s\n",
      "Epoch 145: Accuracy: 20.29117% (223/1099), Higher: 996, Lower: 868, Time: 8.42405s\n",
      "Epoch 146: Accuracy: 20.29117% (223/1099), Higher: 996, Lower: 868, Time: 7.31533s\n",
      "Epoch 147: Accuracy: 20.38217% (224/1099), Higher: 993, Lower: 869, Time: 7.53384s\n",
      "Epoch 148: Accuracy: 20.56415% (226/1099), Higher: 992, Lower: 868, Time: 7.38267s\n",
      "Epoch 149: Accuracy: 20.47316% (225/1099), Higher: 992, Lower: 869, Time: 7.16099s\n",
      "Epoch 150: Accuracy: 20.47316% (225/1099), Higher: 992, Lower: 869, Time: 7.04625s\n",
      "Epoch 151: Accuracy: 20.38217% (224/1099), Higher: 995, Lower: 867, Time: 7.71654s\n",
      "Epoch 152: Accuracy: 20.56415% (226/1099), Higher: 992, Lower: 864, Time: 7.34104s\n",
      "Epoch 153: Accuracy: 20.56415% (226/1099), Higher: 992, Lower: 869, Time: 7.61969s\n",
      "Epoch 154: Accuracy: 20.47316% (225/1099), Higher: 992, Lower: 870, Time: 7.29057s\n",
      "Epoch 155: Accuracy: 20.56415% (226/1099), Higher: 992, Lower: 869, Time: 7.32089s\n",
      "Epoch 156: Accuracy: 20.56415% (226/1099), Higher: 991, Lower: 869, Time: 7.26995s\n",
      "Epoch 157: Accuracy: 20.56415% (226/1099), Higher: 992, Lower: 869, Time: 7.32671s\n",
      "Epoch 158: Accuracy: 20.56415% (226/1099), Higher: 991, Lower: 869, Time: 7.29383s\n",
      "Epoch 159: Accuracy: 20.56415% (226/1099), Higher: 992, Lower: 869, Time: 7.07134s\n",
      "Epoch 160: Accuracy: 20.56415% (226/1099), Higher: 992, Lower: 869, Time: 7.17523s\n",
      "Epoch 161: Accuracy: 20.56415% (226/1099), Higher: 992, Lower: 869, Time: 7.23080s\n",
      "Epoch 162: Accuracy: 20.56415% (226/1099), Higher: 995, Lower: 867, Time: 7.24546s\n",
      "Epoch 163: Accuracy: 20.56415% (226/1099), Higher: 994, Lower: 867, Time: 7.16065s\n",
      "Epoch 164: Accuracy: 20.56415% (226/1099), Higher: 994, Lower: 869, Time: 7.29891s\n",
      "Epoch 165: Accuracy: 20.56415% (226/1099), Higher: 993, Lower: 865, Time: 7.29472s\n",
      "Epoch 166: Accuracy: 20.56415% (226/1099), Higher: 994, Lower: 865, Time: 7.25441s\n",
      "Epoch 167: Accuracy: 20.56415% (226/1099), Higher: 993, Lower: 865, Time: 7.12927s\n",
      "Epoch 168: Accuracy: 20.56415% (226/1099), Higher: 993, Lower: 863, Time: 7.33462s\n",
      "Epoch 169: Accuracy: 20.56415% (226/1099), Higher: 996, Lower: 863, Time: 7.04894s\n",
      "Epoch 170: Accuracy: 20.56415% (226/1099), Higher: 993, Lower: 864, Time: 7.31565s\n",
      "Epoch 171: Accuracy: 20.47316% (225/1099), Higher: 993, Lower: 865, Time: 7.27802s\n",
      "Epoch 172: Accuracy: 20.65514% (227/1099), Higher: 991, Lower: 864, Time: 7.11662s\n",
      "Epoch 173: Accuracy: 20.65514% (227/1099), Higher: 991, Lower: 863, Time: 7.45898s\n",
      "Epoch 174: Accuracy: 20.65514% (227/1099), Higher: 987, Lower: 861, Time: 7.24051s\n",
      "Epoch 175: Accuracy: 20.74613% (228/1099), Higher: 994, Lower: 853, Time: 7.23320s\n",
      "Epoch 176: Accuracy: 20.74613% (228/1099), Higher: 991, Lower: 853, Time: 7.19377s\n",
      "Epoch 177: Accuracy: 20.74613% (228/1099), Higher: 997, Lower: 857, Time: 7.19304s\n",
      "Epoch 178: Accuracy: 20.65514% (227/1099), Higher: 1001, Lower: 858, Time: 7.41966s\n",
      "Epoch 179: Accuracy: 20.65514% (227/1099), Higher: 1001, Lower: 858, Time: 7.19316s\n",
      "Epoch 180: Accuracy: 20.65514% (227/1099), Higher: 998, Lower: 860, Time: 7.27152s\n",
      "Epoch 181: Accuracy: 20.65514% (227/1099), Higher: 1001, Lower: 854, Time: 7.03878s\n",
      "Epoch 182: Accuracy: 20.65514% (227/1099), Higher: 1001, Lower: 854, Time: 7.25408s\n",
      "Epoch 183: Accuracy: 20.74613% (228/1099), Higher: 1001, Lower: 851, Time: 7.16211s\n",
      "Epoch 184: Accuracy: 20.74613% (228/1099), Higher: 999, Lower: 849, Time: 7.28127s\n",
      "Epoch 185: Accuracy: 20.74613% (228/1099), Higher: 1000, Lower: 851, Time: 7.08664s\n",
      "Epoch 186: Accuracy: 20.65514% (227/1099), Higher: 995, Lower: 857, Time: 7.34245s\n",
      "Epoch 187: Accuracy: 20.65514% (227/1099), Higher: 996, Lower: 850, Time: 7.55346s\n",
      "Epoch 188: Accuracy: 20.74613% (228/1099), Higher: 993, Lower: 846, Time: 7.30248s\n",
      "Epoch 189: Accuracy: 20.56415% (226/1099), Higher: 995, Lower: 853, Time: 7.25173s\n",
      "Epoch 190: Accuracy: 20.65514% (227/1099), Higher: 991, Lower: 851, Time: 7.34034s\n",
      "Epoch 191: Accuracy: 20.65514% (227/1099), Higher: 991, Lower: 851, Time: 7.35076s\n",
      "Epoch 192: Accuracy: 20.56415% (226/1099), Higher: 991, Lower: 856, Time: 7.36622s\n",
      "Epoch 193: Accuracy: 20.74613% (228/1099), Higher: 991, Lower: 846, Time: 7.26637s\n",
      "Epoch 194: Accuracy: 20.74613% (228/1099), Higher: 991, Lower: 846, Time: 7.10051s\n",
      "Epoch 195: Accuracy: 20.74613% (228/1099), Higher: 991, Lower: 846, Time: 7.14259s\n",
      "Epoch 196: Accuracy: 20.74613% (228/1099), Higher: 991, Lower: 846, Time: 7.28928s\n",
      "Epoch 197: Accuracy: 20.83712% (229/1099), Higher: 990, Lower: 846, Time: 7.18152s\n",
      "Epoch 198: Accuracy: 20.74613% (228/1099), Higher: 991, Lower: 846, Time: 7.07710s\n",
      "Epoch 199: Accuracy: 20.74613% (228/1099), Higher: 991, Lower: 846, Time: 7.13537s\n"
     ]
    }
   ],
   "execution_count": 9
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:38:46.945595Z",
     "start_time": "2024-05-28T09:38:46.742349Z"
    }
   },
   "cell_type": "code",
   "source": [
    "plt.plot(accuracy)\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.show()"
   ],
   "id": "ca130e6f21c3b2fd",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 10
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### validate",
   "id": "929610fa5827f708"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:38:47.034240Z",
     "start_time": "2024-05-28T09:38:46.948654Z"
    }
   },
   "cell_type": "code",
   "source": [
    "val_set = data.iloc[:, :-2]\n",
    "val_dose = data.iloc[:, -1:].values"
   ],
   "id": "64d9315a8e051563",
   "outputs": [],
   "execution_count": 11
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:38:54.328382Z",
     "start_time": "2024-05-28T09:38:47.035083Z"
    }
   },
   "cell_type": "code",
   "source": [
    "arms = np.arange(12)\n",
    "correct_count = 0\n",
    "count = 0\n",
    "heatmap = np.zeros((12, 12))\n",
    "\n",
    "for i in range(len(data)):\n",
    "    count += 1\n",
    "    feature = np.array(val_set.iloc[i], dtype=float)\n",
    "    # iterate through actions\n",
    "    p = {}\n",
    "    max_value = -np.inf\n",
    "    max_arm = 0\n",
    "    for arm in arms:\n",
    "        theta = np.matmul(np.linalg.inv(A_map[arm]), b_map[arm])\n",
    "        p = np.matmul(theta.T, feature) + \\\n",
    "            np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map[arm])), feature))\n",
    "        if p == max_value:\n",
    "            if np.random.choice(a=[False, True]): max_arm = arm\n",
    "        elif p > max_value:\n",
    "            max_value = p\n",
    "            max_arm = arm\n",
    "\n",
    "    if max_arm == val_dose[i][0]:\n",
    "        correct_count += 1\n",
    "    heatmap[max_arm][int(val_dose[i][0])] += 1\n",
    "\n",
    "accu = correct_count / count\n",
    "print(\"Validate: Accuracy: %.5f%% (%d/%d)\" % (accu * 100, correct_count, count))"
   ],
   "id": "abd2903bf57b57d1",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validate: Accuracy: 27.05433% (991/3663)\n"
     ]
    }
   ],
   "execution_count": 12
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:38:54.691765Z",
     "start_time": "2024-05-28T09:38:54.329458Z"
    }
   },
   "cell_type": "code",
   "source": [
    "ax = sns.heatmap(heatmap, linewidth=0.5, annot=True, cmap='Blues', fmt='.20g')\n",
    "plt.show()"
   ],
   "id": "ff0a3918fd8be9b4",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 13
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": [
    "## Long range\n",
    "### Filter"
   ],
   "id": "de0d291d4dce14a"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:38:54.702148Z",
     "start_time": "2024-05-28T09:38:54.692862Z"
    }
   },
   "cell_type": "code",
   "source": [
    "data1 = data.copy()\n",
    "data1.loc[data1['dose'] < 21,'type'] = 0\n",
    "data1.loc[(data1['dose'] >= 21) & (data1['dose'] < 49),'type'] = 1\n",
    "data1.loc[data1['dose'] >= 49,'type'] = 2"
   ],
   "id": "fea27cf00740e603",
   "outputs": [],
   "execution_count": 14
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Split",
   "id": "35530e43d9e855da"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:38:54.790079Z",
     "start_time": "2024-05-28T09:38:54.703187Z"
    }
   },
   "cell_type": "code",
   "source": [
    "data_size1 = len(data1)\n",
    "ratio1 = 0.7\n",
    "pointer = int(ratio1 * data_size1)\n",
    "train_set1 = data1.iloc[:pointer, :-2]\n",
    "test_set1 = data1.iloc[pointer:, :-2]\n",
    "train_dose1 = data1.iloc[:pointer, -1:].values\n",
    "test_dose1 = data1.iloc[pointer:, -1:].values"
   ],
   "id": "c4d8d06bb7cfbff0",
   "outputs": [],
   "execution_count": 15
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Train",
   "id": "5c20936e04319b56"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:46:00.531292Z",
     "start_time": "2024-05-28T09:38:54.791103Z"
    }
   },
   "cell_type": "code",
   "source": [
    "arms = np.arange(3)\n",
    "accuracy1 = []\n",
    "A_map1 = {}\n",
    "b_map1 = {}\n",
    "for arm in arms:\n",
    "    A_map1[arm] = np.identity(len(list(train_set1)))\n",
    "    b_map1[arm] = np.zeros(len(list(train_set1)))\n",
    "\n",
    "for epoch in range(200):\n",
    "    start_time = time.time()\n",
    "    # train\n",
    "    indices = np.random.permutation(list(range(len(train_set1))))\n",
    "    for i in indices:\n",
    "        feature = np.array(train_set1.iloc[i], dtype=float)\n",
    "        # iterate through actions\n",
    "        p = {}\n",
    "        max_value = -np.inf\n",
    "        max_arm = 0\n",
    "        for arm in arms:\n",
    "            theta = np.matmul(np.linalg.inv(A_map1[arm]), b_map1[arm])\n",
    "            p = np.matmul(theta.T, feature) + \\\n",
    "                np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map1[arm])), feature))\n",
    "\n",
    "            if p == max_value:\n",
    "                if np.random.choice(a=[False, True]): max_arm = arm\n",
    "            elif p > max_value:\n",
    "                max_value = p\n",
    "                max_arm = arm\n",
    "\n",
    "        # observe reward and update A and b\n",
    "        r = -abs(train_dose1[i][0]-max_arm)\n",
    "        A_map1[max_arm] += np.outer(feature, feature)\n",
    "        b_map1[max_arm] += r * feature\n",
    "\n",
    "    # test\n",
    "    correct_count = 0\n",
    "    count = 0\n",
    "    higher = 0\n",
    "    lower = 0\n",
    "    for i in range(len(test_set1)):\n",
    "        count += 1\n",
    "        feature = np.array(test_set1.iloc[i], dtype=float)\n",
    "        # iterate through actions\n",
    "        p = {}\n",
    "        max_value = -np.inf\n",
    "        max_arm = 0\n",
    "        for arm in arms:\n",
    "            theta = np.matmul(np.linalg.inv(A_map1[arm]), b_map1[arm])\n",
    "            p = np.matmul(theta.T, feature) + \\\n",
    "                np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map1[arm])), feature))\n",
    "            if p == max_value:\n",
    "                if np.random.choice(a=[False, True]): max_arm = arm\n",
    "            elif p > max_value:\n",
    "                max_value = p\n",
    "                max_arm = arm\n",
    "\n",
    "        if max_arm == test_dose1[i]:\n",
    "            correct_count += 1\n",
    "        elif max_arm < test_dose1[i]:\n",
    "            lower += test_dose1[i] - max_arm\n",
    "        else:\n",
    "            higher += max_arm - test_dose1[i]\n",
    "\n",
    "    end_time = time.time()\n",
    "    accu = correct_count / count\n",
    "    print(\"Epoch %d: Accuracy: %.5f%% (%d/%d), Higher: %d, Lower: %d, Time: %.5fs\" % (\n",
    "        epoch, accu * 100, correct_count, count, higher, lower, end_time - start_time))\n",
    "    accuracy1.append(accu)"
   ],
   "id": "172615e4c47c351b",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0: Accuracy: 62.42038% (686/1099), Higher: 259, Lower: 169, Time: 2.17541s\n",
      "Epoch 1: Accuracy: 63.96724% (703/1099), Higher: 236, Lower: 169, Time: 2.09345s\n",
      "Epoch 2: Accuracy: 64.60419% (710/1099), Higher: 221, Lower: 174, Time: 2.05585s\n",
      "Epoch 3: Accuracy: 64.69518% (711/1099), Higher: 218, Lower: 175, Time: 2.14996s\n",
      "Epoch 4: Accuracy: 65.33212% (718/1099), Higher: 213, Lower: 172, Time: 2.06784s\n",
      "Epoch 5: Accuracy: 65.15014% (716/1099), Higher: 212, Lower: 176, Time: 2.06511s\n",
      "Epoch 6: Accuracy: 65.05914% (715/1099), Higher: 212, Lower: 177, Time: 2.08781s\n",
      "Epoch 7: Accuracy: 65.15014% (716/1099), Higher: 212, Lower: 176, Time: 2.07655s\n",
      "Epoch 8: Accuracy: 64.96815% (714/1099), Higher: 213, Lower: 177, Time: 2.07675s\n",
      "Epoch 9: Accuracy: 65.42311% (719/1099), Higher: 210, Lower: 175, Time: 2.07365s\n",
      "Epoch 10: Accuracy: 65.24113% (717/1099), Higher: 211, Lower: 176, Time: 2.10106s\n",
      "Epoch 11: Accuracy: 65.24113% (717/1099), Higher: 211, Lower: 176, Time: 2.10075s\n",
      "Epoch 12: Accuracy: 65.05914% (715/1099), Higher: 211, Lower: 178, Time: 2.21502s\n",
      "Epoch 13: Accuracy: 65.15014% (716/1099), Higher: 210, Lower: 178, Time: 2.19862s\n",
      "Epoch 14: Accuracy: 64.96815% (714/1099), Higher: 212, Lower: 178, Time: 2.17622s\n",
      "Epoch 15: Accuracy: 64.96815% (714/1099), Higher: 208, Lower: 182, Time: 2.19644s\n",
      "Epoch 16: Accuracy: 64.96815% (714/1099), Higher: 210, Lower: 180, Time: 2.07429s\n",
      "Epoch 17: Accuracy: 65.15014% (716/1099), Higher: 208, Lower: 180, Time: 2.07125s\n",
      "Epoch 18: Accuracy: 64.87716% (713/1099), Higher: 209, Lower: 182, Time: 2.10897s\n",
      "Epoch 19: Accuracy: 64.87716% (713/1099), Higher: 209, Lower: 182, Time: 2.17794s\n",
      "Epoch 20: Accuracy: 64.96815% (714/1099), Higher: 208, Lower: 182, Time: 2.06632s\n",
      "Epoch 21: Accuracy: 64.96815% (714/1099), Higher: 208, Lower: 182, Time: 2.15660s\n",
      "Epoch 22: Accuracy: 64.78617% (712/1099), Higher: 207, Lower: 185, Time: 2.16455s\n",
      "Epoch 23: Accuracy: 64.87716% (713/1099), Higher: 207, Lower: 184, Time: 2.19668s\n",
      "Epoch 24: Accuracy: 64.78617% (712/1099), Higher: 208, Lower: 184, Time: 2.07906s\n",
      "Epoch 25: Accuracy: 64.69518% (711/1099), Higher: 207, Lower: 186, Time: 2.22732s\n",
      "Epoch 26: Accuracy: 64.87716% (713/1099), Higher: 207, Lower: 184, Time: 2.14349s\n",
      "Epoch 27: Accuracy: 64.69518% (711/1099), Higher: 207, Lower: 186, Time: 2.11968s\n",
      "Epoch 28: Accuracy: 64.96815% (714/1099), Higher: 207, Lower: 183, Time: 2.19172s\n",
      "Epoch 29: Accuracy: 64.87716% (713/1099), Higher: 207, Lower: 184, Time: 2.15504s\n",
      "Epoch 30: Accuracy: 64.78617% (712/1099), Higher: 207, Lower: 185, Time: 2.11830s\n",
      "Epoch 31: Accuracy: 65.05914% (715/1099), Higher: 207, Lower: 182, Time: 2.09875s\n",
      "Epoch 32: Accuracy: 65.05914% (715/1099), Higher: 206, Lower: 183, Time: 2.09059s\n",
      "Epoch 33: Accuracy: 65.05914% (715/1099), Higher: 206, Lower: 183, Time: 2.10522s\n",
      "Epoch 34: Accuracy: 64.78617% (712/1099), Higher: 207, Lower: 185, Time: 2.09440s\n",
      "Epoch 35: Accuracy: 64.96815% (714/1099), Higher: 206, Lower: 184, Time: 2.08434s\n",
      "Epoch 36: Accuracy: 64.87716% (713/1099), Higher: 207, Lower: 184, Time: 2.08128s\n",
      "Epoch 37: Accuracy: 64.87716% (713/1099), Higher: 206, Lower: 185, Time: 2.16716s\n",
      "Epoch 38: Accuracy: 65.15014% (716/1099), Higher: 204, Lower: 184, Time: 2.21994s\n",
      "Epoch 39: Accuracy: 65.15014% (716/1099), Higher: 204, Lower: 184, Time: 2.19585s\n",
      "Epoch 40: Accuracy: 64.96815% (714/1099), Higher: 204, Lower: 186, Time: 2.22381s\n",
      "Epoch 41: Accuracy: 65.05914% (715/1099), Higher: 204, Lower: 185, Time: 2.21410s\n",
      "Epoch 42: Accuracy: 64.87716% (713/1099), Higher: 205, Lower: 186, Time: 2.15431s\n",
      "Epoch 43: Accuracy: 64.96815% (714/1099), Higher: 204, Lower: 186, Time: 2.17952s\n",
      "Epoch 44: Accuracy: 64.96815% (714/1099), Higher: 204, Lower: 186, Time: 2.06425s\n",
      "Epoch 45: Accuracy: 64.96815% (714/1099), Higher: 204, Lower: 186, Time: 2.07911s\n",
      "Epoch 46: Accuracy: 64.87716% (713/1099), Higher: 205, Lower: 186, Time: 2.08212s\n",
      "Epoch 47: Accuracy: 64.87716% (713/1099), Higher: 205, Lower: 186, Time: 2.08140s\n",
      "Epoch 48: Accuracy: 64.96815% (714/1099), Higher: 204, Lower: 186, Time: 2.09096s\n",
      "Epoch 49: Accuracy: 64.96815% (714/1099), Higher: 204, Lower: 186, Time: 2.10483s\n",
      "Epoch 50: Accuracy: 64.87716% (713/1099), Higher: 204, Lower: 187, Time: 2.19275s\n",
      "Epoch 51: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 187, Time: 2.08128s\n",
      "Epoch 52: Accuracy: 64.87716% (713/1099), Higher: 204, Lower: 187, Time: 2.19772s\n",
      "Epoch 53: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 187, Time: 2.19973s\n",
      "Epoch 54: Accuracy: 64.87716% (713/1099), Higher: 204, Lower: 187, Time: 2.18276s\n",
      "Epoch 55: Accuracy: 64.87716% (713/1099), Higher: 204, Lower: 187, Time: 2.08485s\n",
      "Epoch 56: Accuracy: 64.87716% (713/1099), Higher: 204, Lower: 187, Time: 2.08082s\n",
      "Epoch 57: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 187, Time: 2.09005s\n",
      "Epoch 58: Accuracy: 64.69518% (711/1099), Higher: 205, Lower: 188, Time: 2.09134s\n",
      "Epoch 59: Accuracy: 64.69518% (711/1099), Higher: 205, Lower: 187, Time: 2.10909s\n",
      "Epoch 60: Accuracy: 64.69518% (711/1099), Higher: 205, Lower: 188, Time: 2.08995s\n",
      "Epoch 61: Accuracy: 64.69518% (711/1099), Higher: 205, Lower: 187, Time: 2.11704s\n",
      "Epoch 62: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 186, Time: 2.08882s\n",
      "Epoch 63: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 186, Time: 2.09384s\n",
      "Epoch 64: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 186, Time: 2.15154s\n",
      "Epoch 65: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 186, Time: 2.18331s\n",
      "Epoch 66: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 186, Time: 2.16811s\n",
      "Epoch 67: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 186, Time: 2.16799s\n",
      "Epoch 68: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 186, Time: 2.13261s\n",
      "Epoch 69: Accuracy: 64.78617% (712/1099), Higher: 205, Lower: 186, Time: 2.14969s\n",
      "Epoch 70: Accuracy: 64.69518% (711/1099), Higher: 205, Lower: 187, Time: 2.08935s\n",
      "Epoch 71: Accuracy: 64.69518% (711/1099), Higher: 204, Lower: 188, Time: 2.10291s\n",
      "Epoch 72: Accuracy: 64.69518% (711/1099), Higher: 204, Lower: 188, Time: 2.09400s\n",
      "Epoch 73: Accuracy: 64.69518% (711/1099), Higher: 204, Lower: 188, Time: 2.13106s\n",
      "Epoch 74: Accuracy: 64.87716% (713/1099), Higher: 203, Lower: 187, Time: 2.08414s\n",
      "Epoch 75: Accuracy: 64.96815% (714/1099), Higher: 202, Lower: 187, Time: 2.09396s\n",
      "Epoch 76: Accuracy: 64.96815% (714/1099), Higher: 202, Lower: 187, Time: 2.11508s\n",
      "Epoch 77: Accuracy: 64.96815% (714/1099), Higher: 202, Lower: 187, Time: 2.10819s\n",
      "Epoch 78: Accuracy: 64.96815% (714/1099), Higher: 202, Lower: 187, Time: 2.14699s\n",
      "Epoch 79: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.08904s\n",
      "Epoch 80: Accuracy: 64.96815% (714/1099), Higher: 202, Lower: 187, Time: 2.23072s\n",
      "Epoch 81: Accuracy: 64.87716% (713/1099), Higher: 202, Lower: 188, Time: 2.14855s\n",
      "Epoch 82: Accuracy: 64.96815% (714/1099), Higher: 202, Lower: 187, Time: 2.09721s\n",
      "Epoch 83: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.10085s\n",
      "Epoch 84: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.10468s\n",
      "Epoch 85: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.10538s\n",
      "Epoch 86: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.09724s\n",
      "Epoch 87: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.09019s\n",
      "Epoch 88: Accuracy: 64.96815% (714/1099), Higher: 202, Lower: 187, Time: 2.09099s\n",
      "Epoch 89: Accuracy: 64.96815% (714/1099), Higher: 202, Lower: 187, Time: 2.12245s\n",
      "Epoch 90: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.09171s\n",
      "Epoch 91: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.14810s\n",
      "Epoch 92: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.10538s\n",
      "Epoch 93: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.13368s\n",
      "Epoch 94: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.19774s\n",
      "Epoch 95: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.08541s\n",
      "Epoch 96: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.09071s\n",
      "Epoch 97: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.11931s\n",
      "Epoch 98: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.16256s\n",
      "Epoch 99: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.10935s\n",
      "Epoch 100: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.11287s\n",
      "Epoch 101: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 186, Time: 2.12085s\n",
      "Epoch 102: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.11737s\n",
      "Epoch 103: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.16128s\n",
      "Epoch 104: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.12406s\n",
      "Epoch 105: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.14315s\n",
      "Epoch 106: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.17098s\n",
      "Epoch 107: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.12543s\n",
      "Epoch 108: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.12551s\n",
      "Epoch 109: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.08727s\n",
      "Epoch 110: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 185, Time: 2.10320s\n",
      "Epoch 111: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.08948s\n",
      "Epoch 112: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 185, Time: 2.10675s\n",
      "Epoch 113: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 187, Time: 2.09747s\n",
      "Epoch 114: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 187, Time: 2.20157s\n",
      "Epoch 115: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 187, Time: 2.10169s\n",
      "Epoch 116: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 187, Time: 2.09891s\n",
      "Epoch 117: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 187, Time: 2.08524s\n",
      "Epoch 118: Accuracy: 64.96815% (714/1099), Higher: 203, Lower: 187, Time: 2.09571s\n",
      "Epoch 119: Accuracy: 64.96815% (714/1099), Higher: 203, Lower: 187, Time: 2.10251s\n",
      "Epoch 120: Accuracy: 64.96815% (714/1099), Higher: 203, Lower: 187, Time: 2.11599s\n",
      "Epoch 121: Accuracy: 64.96815% (714/1099), Higher: 203, Lower: 187, Time: 2.12328s\n",
      "Epoch 122: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 187, Time: 2.12972s\n",
      "Epoch 123: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 187, Time: 2.13368s\n",
      "Epoch 124: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 187, Time: 2.16964s\n",
      "Epoch 125: Accuracy: 65.05914% (715/1099), Higher: 202, Lower: 187, Time: 2.24690s\n",
      "Epoch 126: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.13176s\n",
      "Epoch 127: Accuracy: 64.96815% (714/1099), Higher: 203, Lower: 187, Time: 2.15212s\n",
      "Epoch 128: Accuracy: 64.96815% (714/1099), Higher: 203, Lower: 187, Time: 2.12056s\n",
      "Epoch 129: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.14663s\n",
      "Epoch 130: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.14412s\n",
      "Epoch 131: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.13744s\n",
      "Epoch 132: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.15716s\n",
      "Epoch 133: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.13017s\n",
      "Epoch 134: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.17040s\n",
      "Epoch 135: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.09175s\n",
      "Epoch 136: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.10227s\n",
      "Epoch 137: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.10957s\n",
      "Epoch 138: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.10439s\n",
      "Epoch 139: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.13530s\n",
      "Epoch 140: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.13296s\n",
      "Epoch 141: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.10845s\n",
      "Epoch 142: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.10855s\n",
      "Epoch 143: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.10214s\n",
      "Epoch 144: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.10426s\n",
      "Epoch 145: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.10058s\n",
      "Epoch 146: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.15139s\n",
      "Epoch 147: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.11651s\n",
      "Epoch 148: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.11469s\n",
      "Epoch 149: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.09807s\n",
      "Epoch 150: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.08839s\n",
      "Epoch 151: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.09153s\n",
      "Epoch 152: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.17742s\n",
      "Epoch 153: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.19336s\n",
      "Epoch 154: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.14040s\n",
      "Epoch 155: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.11071s\n",
      "Epoch 156: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.12528s\n",
      "Epoch 157: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.11194s\n",
      "Epoch 158: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.13363s\n",
      "Epoch 159: Accuracy: 65.05914% (715/1099), Higher: 203, Lower: 186, Time: 2.09760s\n",
      "Epoch 160: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.08656s\n",
      "Epoch 161: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.08603s\n",
      "Epoch 162: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.12193s\n",
      "Epoch 163: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.12085s\n",
      "Epoch 164: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.09240s\n",
      "Epoch 165: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.10432s\n",
      "Epoch 166: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.09400s\n",
      "Epoch 167: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.10487s\n",
      "Epoch 168: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.11905s\n",
      "Epoch 169: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.13467s\n",
      "Epoch 170: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.16122s\n",
      "Epoch 171: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.16065s\n",
      "Epoch 172: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.11303s\n",
      "Epoch 173: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.10212s\n",
      "Epoch 174: Accuracy: 65.24113% (717/1099), Higher: 202, Lower: 185, Time: 2.20688s\n",
      "Epoch 175: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.17206s\n",
      "Epoch 176: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.14147s\n",
      "Epoch 177: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.11881s\n",
      "Epoch 178: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.11531s\n",
      "Epoch 179: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.14571s\n",
      "Epoch 180: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.12466s\n",
      "Epoch 181: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.20666s\n",
      "Epoch 182: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.22845s\n",
      "Epoch 183: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.17908s\n",
      "Epoch 184: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.20580s\n",
      "Epoch 185: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.16916s\n",
      "Epoch 186: Accuracy: 65.24113% (717/1099), Higher: 201, Lower: 186, Time: 2.18760s\n",
      "Epoch 187: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.14088s\n",
      "Epoch 188: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.13699s\n",
      "Epoch 189: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.11445s\n",
      "Epoch 190: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.11225s\n",
      "Epoch 191: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.11556s\n",
      "Epoch 192: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.16437s\n",
      "Epoch 193: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.10165s\n",
      "Epoch 194: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.09959s\n",
      "Epoch 195: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.11384s\n",
      "Epoch 196: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.11200s\n",
      "Epoch 197: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.14361s\n",
      "Epoch 198: Accuracy: 65.24113% (717/1099), Higher: 201, Lower: 186, Time: 2.23491s\n",
      "Epoch 199: Accuracy: 65.15014% (716/1099), Higher: 202, Lower: 186, Time: 2.19065s\n"
     ]
    }
   ],
   "execution_count": 16
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:46:00.678939Z",
     "start_time": "2024-05-28T09:46:00.532301Z"
    }
   },
   "cell_type": "code",
   "source": [
    "plt.plot(accuracy1)\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.show()"
   ],
   "id": "50432ff853b110fb",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 17
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Validate",
   "id": "315fdb65cdaf527a"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:46:00.683794Z",
     "start_time": "2024-05-28T09:46:00.679937Z"
    }
   },
   "cell_type": "code",
   "source": [
    "val_set1 = data1.iloc[:, :-2]\n",
    "val_dose1 = data1.iloc[:, -1:].values"
   ],
   "id": "97311b343b038520",
   "outputs": [],
   "execution_count": 18
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:46:02.785009Z",
     "start_time": "2024-05-28T09:46:00.684742Z"
    }
   },
   "cell_type": "code",
   "source": [
    "arms = np.arange(3)\n",
    "correct_count = 0\n",
    "count = 0\n",
    "heatmap1 = np.zeros((3, 3))\n",
    "\n",
    "for i in range(len(data1)):\n",
    "    count += 1\n",
    "    feature = np.array(val_set1.iloc[i], dtype=float)\n",
    "    # iterate through actions\n",
    "    p = {}\n",
    "    max_value = -np.inf\n",
    "    max_arm = 0\n",
    "    for arm in arms:\n",
    "        theta = np.matmul(np.linalg.inv(A_map1[arm]), b_map1[arm])\n",
    "        p = np.matmul(theta.T, feature) + \\\n",
    "            np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map1[arm])), feature))\n",
    "        if p == max_value:\n",
    "            if np.random.choice(a=[False, True]): max_arm = arm\n",
    "        elif p > max_value:\n",
    "            max_value = p\n",
    "            max_arm = arm\n",
    "\n",
    "    if max_arm == val_dose1[i][0]:\n",
    "        correct_count += 1\n",
    "    heatmap1[max_arm][int(val_dose1[i][0])] += 1\n",
    "\n",
    "accu = correct_count / count\n",
    "print(\"Validate: Accuracy: %.5f%% (%d/%d)\" % (accu * 100, correct_count, count))"
   ],
   "id": "fdc0000aa8ef174b",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validate: Accuracy: 71.22577% (2609/3663)\n"
     ]
    }
   ],
   "execution_count": 19
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:46:02.929804Z",
     "start_time": "2024-05-28T09:46:02.785991Z"
    }
   },
   "cell_type": "code",
   "source": [
    "ax = sns.heatmap(heatmap1, linewidth=0.5, annot=True, cmap='Blues', fmt='.20g')\n",
    "plt.show()"
   ],
   "id": "a4dda1dd5f0dfdf4",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 20
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": [
    "# Train and test on real data with threshold\n",
    "## Short range\n",
    "### Filter"
   ],
   "id": "81636dd767e04bc1"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:46:02.949457Z",
     "start_time": "2024-05-28T09:46:02.930650Z"
    }
   },
   "cell_type": "code",
   "source": [
    "data2 = data.drop(data.columns[data.sum() < data_size*0.05], axis=1)\n",
    "data2.info()"
   ],
   "id": "97f1efcc63865574",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 3663 entries, 0 to 3662\n",
      "Data columns (total 36 columns):\n",
      " #   Column                                          Non-Null Count  Dtype  \n",
      "---  ------                                          --------------  -----  \n",
      " 0   Gender                                          3663 non-null   object \n",
      " 1   Age in decades                                  3663 non-null   float64\n",
      " 2   Height (cm)                                     3663 non-null   float64\n",
      " 3   Weight (kg)                                     3663 non-null   float64\n",
      " 4   Race Asian                                      3663 non-null   object \n",
      " 5   Race Black or African American                  3663 non-null   object \n",
      " 6   Race Caucasian or White                         3663 non-null   object \n",
      " 7   Race Unknown                                    3663 non-null   object \n",
      " 8   DVT                                             3663 non-null   object \n",
      " 9   PE                                              3663 non-null   object \n",
      " 10  Afib/flutter                                    3663 non-null   object \n",
      " 11  Heart Valve                                     3663 non-null   object \n",
      " 12  Other                                           3663 non-null   object \n",
      " 13  Diabetes                                        3663 non-null   float64\n",
      " 14  Congestive Heart Failure and/or Cardiomyopathy  3663 non-null   float64\n",
      " 15  Valve Replacement                               3663 non-null   float64\n",
      " 16  Aspirin                                         3663 non-null   float64\n",
      " 17  Acetaminophen or Paracetamol (Tylenol)          3663 non-null   float64\n",
      " 18  Simvastatin (Zocor)                             3663 non-null   float64\n",
      " 19  Atorvastatin (Lipitor)                          3663 non-null   float64\n",
      " 20  Amiodarone (Cordarone)                          3663 non-null   float64\n",
      " 21  Herbal Medications, Vitamins, Supplements       3663 non-null   float64\n",
      " 22  Current Smoker                                  3663 non-null   float64\n",
      " 23  CYP2C9 *1                                       3663 non-null   object \n",
      " 24  CYP2C9 *2                                       3663 non-null   object \n",
      " 25  CYP2C9 *3                                       3663 non-null   object \n",
      " 26  VKORC1 - -1639G>A                               3663 non-null   object \n",
      " 27  VKORC1 - 497T>G                                 3663 non-null   object \n",
      " 28  VKORC1 - 1173C>T                                3663 non-null   object \n",
      " 29  VKORC1 - 1542G>C                                3663 non-null   object \n",
      " 30  VKORC1 - 3730G>A                                3663 non-null   object \n",
      " 31  VKORC1 - 2255C>T                                3663 non-null   object \n",
      " 32  VKORC1 - -4451C>A                               3663 non-null   object \n",
      " 33  Target INR                                      3663 non-null   float64\n",
      " 34  dose                                            3663 non-null   float64\n",
      " 35  type                                            3663 non-null   object \n",
      "dtypes: float64(15), object(21)\n",
      "memory usage: 1.0+ MB\n"
     ]
    }
   ],
   "execution_count": 21
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Split",
   "id": "f0186ef336994da2"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:46:03.013285Z",
     "start_time": "2024-05-28T09:46:02.950672Z"
    }
   },
   "cell_type": "code",
   "source": [
    "data_size2 = len(data2)\n",
    "ratio2 = 0.7\n",
    "pointer = int(ratio2 * data_size2)\n",
    "train_set2 = data2.iloc[:pointer, :-2]\n",
    "test_set2 = data2.iloc[pointer:, :-2]\n",
    "train_dose2 = data2.iloc[:pointer, -1:].values\n",
    "test_dose2 = data2.iloc[pointer:, -1:].values"
   ],
   "id": "b8fb63a1d8794982",
   "outputs": [],
   "execution_count": 22
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Train",
   "id": "73040c8dea27e3bd"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:55:52.597653Z",
     "start_time": "2024-05-28T09:46:03.014340Z"
    }
   },
   "cell_type": "code",
   "source": [
    "arms = np.arange(12)\n",
    "accuracy2 = []\n",
    "A_map2 = {}\n",
    "b_map2 = {}\n",
    "for arm in arms:\n",
    "    A_map2[arm] = np.identity(len(list(train_set2)))\n",
    "    b_map2[arm] = np.zeros(len(list(train_set2)))\n",
    "\n",
    "for epoch in range(200):\n",
    "    start_time = time.time()\n",
    "    # train\n",
    "    indices = np.random.permutation(list(range(len(train_set2))))\n",
    "    for i in indices:\n",
    "        feature = np.array(train_set2.iloc[i], dtype=float)\n",
    "        # iterate through actions\n",
    "        p = {}\n",
    "        max_value = -np.inf\n",
    "        max_arm = 0\n",
    "        for arm in arms:\n",
    "            theta = np.matmul(np.linalg.inv(A_map2[arm]), b_map2[arm])\n",
    "            p = np.matmul(theta.T, feature) + \\\n",
    "                np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map2[arm])), feature))\n",
    "\n",
    "            if p == max_value:\n",
    "                if np.random.choice(a=[False, True]): max_arm = arm\n",
    "            elif p > max_value:\n",
    "                max_value = p\n",
    "                max_arm = arm\n",
    "\n",
    "        # observe reward and update A and b\n",
    "        r = -abs(train_dose2[i][0]-max_arm)\n",
    "        A_map2[max_arm] += np.outer(feature, feature)\n",
    "        b_map2[max_arm] += r * feature\n",
    "\n",
    "    # test\n",
    "    correct_count = 0\n",
    "    count = 0\n",
    "    higher = 0\n",
    "    lower = 0\n",
    "    for i in range(len(test_set2)):\n",
    "        count += 1\n",
    "        feature = np.array(test_set2.iloc[i], dtype=float)\n",
    "        # iterate through actions\n",
    "        p = {}\n",
    "        max_value = -np.inf\n",
    "        max_arm = 0\n",
    "        for arm in arms:\n",
    "            theta = np.matmul(np.linalg.inv(A_map2[arm]), b_map2[arm])\n",
    "            p = np.matmul(theta.T, feature) + \\\n",
    "                np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map2[arm])), feature))\n",
    "            if p == max_value:\n",
    "                if np.random.choice(a=[False, True]): max_arm = arm\n",
    "            elif p > max_value:\n",
    "                max_value = p\n",
    "                max_arm = arm\n",
    "\n",
    "        if max_arm == test_dose2[i]:\n",
    "            correct_count += 1\n",
    "        elif max_arm < test_dose2[i]:\n",
    "            lower += test_dose2[i] - max_arm\n",
    "        else:\n",
    "            higher += max_arm - test_dose2[i]\n",
    "\n",
    "    end_time = time.time()\n",
    "    accu = correct_count / count\n",
    "    print(\"Epoch %d: Accuracy: %.5f%% (%d/%d), Higher: %d, Lower: %d, Time: %.5fs\" % (\n",
    "        epoch, accu * 100, correct_count, count, higher, lower, end_time - start_time))\n",
    "    accuracy2.append(accu)"
   ],
   "id": "d87bb875dc54bf54",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0: Accuracy: 19.83621% (218/1099), Higher: 936, Lower: 952, Time: 2.90121s\n",
      "Epoch 1: Accuracy: 19.74522% (217/1099), Higher: 792, Lower: 1005, Time: 2.97089s\n",
      "Epoch 2: Accuracy: 18.83530% (207/1099), Higher: 895, Lower: 978, Time: 2.85612s\n",
      "Epoch 3: Accuracy: 19.56324% (215/1099), Higher: 825, Lower: 1037, Time: 2.90085s\n",
      "Epoch 4: Accuracy: 19.10828% (210/1099), Higher: 806, Lower: 1082, Time: 2.88125s\n",
      "Epoch 5: Accuracy: 18.74431% (206/1099), Higher: 831, Lower: 1071, Time: 2.82550s\n",
      "Epoch 6: Accuracy: 18.19836% (200/1099), Higher: 844, Lower: 1062, Time: 2.90699s\n",
      "Epoch 7: Accuracy: 18.65332% (205/1099), Higher: 847, Lower: 1049, Time: 2.81309s\n",
      "Epoch 8: Accuracy: 17.92539% (197/1099), Higher: 851, Lower: 1058, Time: 2.82534s\n",
      "Epoch 9: Accuracy: 18.01638% (198/1099), Higher: 766, Lower: 1108, Time: 2.82929s\n",
      "Epoch 10: Accuracy: 18.01638% (198/1099), Higher: 789, Lower: 1100, Time: 2.83483s\n",
      "Epoch 11: Accuracy: 18.28935% (201/1099), Higher: 783, Lower: 1093, Time: 2.83303s\n",
      "Epoch 12: Accuracy: 18.47134% (203/1099), Higher: 775, Lower: 1097, Time: 2.83862s\n",
      "Epoch 13: Accuracy: 18.28935% (201/1099), Higher: 786, Lower: 1109, Time: 2.83086s\n",
      "Epoch 14: Accuracy: 18.10737% (199/1099), Higher: 787, Lower: 1105, Time: 2.83105s\n",
      "Epoch 15: Accuracy: 17.83439% (196/1099), Higher: 791, Lower: 1102, Time: 2.84498s\n",
      "Epoch 16: Accuracy: 17.74340% (195/1099), Higher: 787, Lower: 1116, Time: 2.88016s\n",
      "Epoch 17: Accuracy: 17.92539% (197/1099), Higher: 788, Lower: 1119, Time: 2.87842s\n",
      "Epoch 18: Accuracy: 18.28935% (201/1099), Higher: 782, Lower: 1116, Time: 2.89397s\n",
      "Epoch 19: Accuracy: 18.47134% (203/1099), Higher: 779, Lower: 1113, Time: 2.93093s\n",
      "Epoch 20: Accuracy: 18.56233% (204/1099), Higher: 777, Lower: 1117, Time: 2.90562s\n",
      "Epoch 21: Accuracy: 18.47134% (203/1099), Higher: 782, Lower: 1124, Time: 2.91025s\n",
      "Epoch 22: Accuracy: 18.38035% (202/1099), Higher: 786, Lower: 1123, Time: 2.91066s\n",
      "Epoch 23: Accuracy: 18.56233% (204/1099), Higher: 793, Lower: 1110, Time: 2.91414s\n",
      "Epoch 24: Accuracy: 18.47134% (203/1099), Higher: 805, Lower: 1113, Time: 2.91050s\n",
      "Epoch 25: Accuracy: 18.47134% (203/1099), Higher: 799, Lower: 1108, Time: 2.88148s\n",
      "Epoch 26: Accuracy: 18.38035% (202/1099), Higher: 792, Lower: 1117, Time: 2.87059s\n",
      "Epoch 27: Accuracy: 18.38035% (202/1099), Higher: 789, Lower: 1117, Time: 2.86186s\n",
      "Epoch 28: Accuracy: 18.56233% (204/1099), Higher: 797, Lower: 1109, Time: 2.86481s\n",
      "Epoch 29: Accuracy: 18.47134% (203/1099), Higher: 795, Lower: 1117, Time: 2.92042s\n",
      "Epoch 30: Accuracy: 18.38035% (202/1099), Higher: 799, Lower: 1110, Time: 3.01874s\n",
      "Epoch 31: Accuracy: 17.74340% (195/1099), Higher: 789, Lower: 1147, Time: 2.88481s\n",
      "Epoch 32: Accuracy: 17.56142% (193/1099), Higher: 769, Lower: 1197, Time: 2.90605s\n",
      "Epoch 33: Accuracy: 17.74340% (195/1099), Higher: 805, Lower: 1104, Time: 2.88390s\n",
      "Epoch 34: Accuracy: 17.92539% (197/1099), Higher: 803, Lower: 1097, Time: 3.07565s\n",
      "Epoch 35: Accuracy: 17.83439% (196/1099), Higher: 811, Lower: 1092, Time: 3.02359s\n",
      "Epoch 36: Accuracy: 17.83439% (196/1099), Higher: 808, Lower: 1094, Time: 2.88283s\n",
      "Epoch 37: Accuracy: 17.83439% (196/1099), Higher: 806, Lower: 1094, Time: 2.89087s\n",
      "Epoch 38: Accuracy: 17.74340% (195/1099), Higher: 814, Lower: 1093, Time: 2.89772s\n",
      "Epoch 39: Accuracy: 17.74340% (195/1099), Higher: 808, Lower: 1095, Time: 2.89637s\n",
      "Epoch 40: Accuracy: 17.74340% (195/1099), Higher: 803, Lower: 1099, Time: 2.89820s\n",
      "Epoch 41: Accuracy: 17.92539% (197/1099), Higher: 805, Lower: 1096, Time: 2.90132s\n",
      "Epoch 42: Accuracy: 17.92539% (197/1099), Higher: 807, Lower: 1093, Time: 2.91075s\n",
      "Epoch 43: Accuracy: 17.83439% (196/1099), Higher: 811, Lower: 1095, Time: 2.89464s\n",
      "Epoch 44: Accuracy: 17.74340% (195/1099), Higher: 809, Lower: 1097, Time: 2.89145s\n",
      "Epoch 45: Accuracy: 17.74340% (195/1099), Higher: 809, Lower: 1098, Time: 2.99890s\n",
      "Epoch 46: Accuracy: 17.74340% (195/1099), Higher: 814, Lower: 1094, Time: 2.95578s\n",
      "Epoch 47: Accuracy: 17.74340% (195/1099), Higher: 814, Lower: 1094, Time: 2.89318s\n",
      "Epoch 48: Accuracy: 17.74340% (195/1099), Higher: 812, Lower: 1094, Time: 2.90315s\n",
      "Epoch 49: Accuracy: 17.65241% (194/1099), Higher: 814, Lower: 1092, Time: 2.96200s\n",
      "Epoch 50: Accuracy: 17.37944% (191/1099), Higher: 816, Lower: 1098, Time: 3.02344s\n",
      "Epoch 51: Accuracy: 17.47043% (192/1099), Higher: 815, Lower: 1094, Time: 2.94333s\n",
      "Epoch 52: Accuracy: 18.38035% (202/1099), Higher: 825, Lower: 1039, Time: 2.96748s\n",
      "Epoch 53: Accuracy: 18.19836% (200/1099), Higher: 833, Lower: 1035, Time: 2.93407s\n",
      "Epoch 54: Accuracy: 18.19836% (200/1099), Higher: 834, Lower: 1035, Time: 2.88895s\n",
      "Epoch 55: Accuracy: 18.01638% (198/1099), Higher: 835, Lower: 1041, Time: 3.09495s\n",
      "Epoch 56: Accuracy: 18.38035% (202/1099), Higher: 834, Lower: 1035, Time: 3.01576s\n",
      "Epoch 57: Accuracy: 18.10737% (199/1099), Higher: 837, Lower: 1043, Time: 2.93692s\n",
      "Epoch 58: Accuracy: 17.92539% (197/1099), Higher: 839, Lower: 1046, Time: 3.01896s\n",
      "Epoch 59: Accuracy: 18.10737% (199/1099), Higher: 843, Lower: 1040, Time: 3.08837s\n",
      "Epoch 60: Accuracy: 18.10737% (199/1099), Higher: 845, Lower: 1041, Time: 3.07762s\n",
      "Epoch 61: Accuracy: 17.92539% (197/1099), Higher: 852, Lower: 1040, Time: 2.96911s\n",
      "Epoch 62: Accuracy: 18.01638% (198/1099), Higher: 851, Lower: 1039, Time: 3.06053s\n",
      "Epoch 63: Accuracy: 17.92539% (197/1099), Higher: 851, Lower: 1041, Time: 2.95861s\n",
      "Epoch 64: Accuracy: 17.83439% (196/1099), Higher: 851, Lower: 1044, Time: 3.03135s\n",
      "Epoch 65: Accuracy: 17.83439% (196/1099), Higher: 856, Lower: 1044, Time: 2.95783s\n",
      "Epoch 66: Accuracy: 17.92539% (197/1099), Higher: 853, Lower: 1048, Time: 2.92904s\n",
      "Epoch 67: Accuracy: 17.92539% (197/1099), Higher: 839, Lower: 1048, Time: 2.94014s\n",
      "Epoch 68: Accuracy: 17.74340% (195/1099), Higher: 836, Lower: 1053, Time: 2.97388s\n",
      "Epoch 69: Accuracy: 17.74340% (195/1099), Higher: 838, Lower: 1052, Time: 2.94831s\n",
      "Epoch 70: Accuracy: 17.83439% (196/1099), Higher: 838, Lower: 1049, Time: 2.93178s\n",
      "Epoch 71: Accuracy: 17.92539% (197/1099), Higher: 841, Lower: 1043, Time: 2.95614s\n",
      "Epoch 72: Accuracy: 18.01638% (198/1099), Higher: 842, Lower: 1038, Time: 2.91711s\n",
      "Epoch 73: Accuracy: 18.01638% (198/1099), Higher: 845, Lower: 1035, Time: 2.98250s\n",
      "Epoch 74: Accuracy: 18.01638% (198/1099), Higher: 847, Lower: 1035, Time: 2.92362s\n",
      "Epoch 75: Accuracy: 18.01638% (198/1099), Higher: 847, Lower: 1032, Time: 3.07245s\n",
      "Epoch 76: Accuracy: 18.01638% (198/1099), Higher: 845, Lower: 1030, Time: 2.89958s\n",
      "Epoch 77: Accuracy: 17.92539% (197/1099), Higher: 845, Lower: 1032, Time: 3.01850s\n",
      "Epoch 78: Accuracy: 18.01638% (198/1099), Higher: 845, Lower: 1031, Time: 3.07707s\n",
      "Epoch 79: Accuracy: 18.01638% (198/1099), Higher: 847, Lower: 1031, Time: 3.09823s\n",
      "Epoch 80: Accuracy: 18.01638% (198/1099), Higher: 848, Lower: 1032, Time: 3.08266s\n",
      "Epoch 81: Accuracy: 18.01638% (198/1099), Higher: 849, Lower: 1029, Time: 3.10588s\n",
      "Epoch 82: Accuracy: 17.92539% (197/1099), Higher: 855, Lower: 1033, Time: 3.07002s\n",
      "Epoch 83: Accuracy: 18.01638% (198/1099), Higher: 848, Lower: 1035, Time: 3.02072s\n",
      "Epoch 84: Accuracy: 18.01638% (198/1099), Higher: 850, Lower: 1031, Time: 2.89296s\n",
      "Epoch 85: Accuracy: 18.10737% (199/1099), Higher: 847, Lower: 1029, Time: 3.03629s\n",
      "Epoch 86: Accuracy: 18.10737% (199/1099), Higher: 847, Lower: 1029, Time: 3.07383s\n",
      "Epoch 87: Accuracy: 17.83439% (196/1099), Higher: 861, Lower: 1020, Time: 2.92029s\n",
      "Epoch 88: Accuracy: 17.83439% (196/1099), Higher: 861, Lower: 1018, Time: 2.97630s\n",
      "Epoch 89: Accuracy: 17.83439% (196/1099), Higher: 865, Lower: 1019, Time: 2.96698s\n",
      "Epoch 90: Accuracy: 17.74340% (195/1099), Higher: 884, Lower: 1018, Time: 3.03674s\n",
      "Epoch 91: Accuracy: 17.74340% (195/1099), Higher: 887, Lower: 1018, Time: 2.89891s\n",
      "Epoch 92: Accuracy: 17.74340% (195/1099), Higher: 886, Lower: 1016, Time: 2.97866s\n",
      "Epoch 93: Accuracy: 17.65241% (194/1099), Higher: 891, Lower: 1016, Time: 3.03783s\n",
      "Epoch 94: Accuracy: 17.65241% (194/1099), Higher: 892, Lower: 1017, Time: 2.96840s\n",
      "Epoch 95: Accuracy: 17.65241% (194/1099), Higher: 889, Lower: 1021, Time: 2.90871s\n",
      "Epoch 96: Accuracy: 17.74340% (195/1099), Higher: 889, Lower: 1019, Time: 2.88968s\n",
      "Epoch 97: Accuracy: 17.74340% (195/1099), Higher: 893, Lower: 1017, Time: 2.90735s\n",
      "Epoch 98: Accuracy: 17.65241% (194/1099), Higher: 894, Lower: 1011, Time: 2.91049s\n",
      "Epoch 99: Accuracy: 17.74340% (195/1099), Higher: 890, Lower: 1013, Time: 2.94407s\n",
      "Epoch 100: Accuracy: 17.83439% (196/1099), Higher: 881, Lower: 1003, Time: 2.98647s\n",
      "Epoch 101: Accuracy: 18.01638% (198/1099), Higher: 880, Lower: 999, Time: 3.08064s\n",
      "Epoch 102: Accuracy: 18.19836% (200/1099), Higher: 882, Lower: 1001, Time: 2.97315s\n",
      "Epoch 103: Accuracy: 18.19836% (200/1099), Higher: 880, Lower: 1001, Time: 3.00494s\n",
      "Epoch 104: Accuracy: 18.47134% (203/1099), Higher: 879, Lower: 996, Time: 2.91839s\n",
      "Epoch 105: Accuracy: 18.47134% (203/1099), Higher: 869, Lower: 997, Time: 2.92428s\n",
      "Epoch 106: Accuracy: 18.47134% (203/1099), Higher: 873, Lower: 1000, Time: 2.91607s\n",
      "Epoch 107: Accuracy: 18.56233% (204/1099), Higher: 870, Lower: 1000, Time: 2.90934s\n",
      "Epoch 108: Accuracy: 18.56233% (204/1099), Higher: 867, Lower: 1001, Time: 2.87016s\n",
      "Epoch 109: Accuracy: 18.65332% (205/1099), Higher: 863, Lower: 1010, Time: 2.89755s\n",
      "Epoch 110: Accuracy: 18.74431% (206/1099), Higher: 859, Lower: 1008, Time: 2.88865s\n",
      "Epoch 111: Accuracy: 18.56233% (204/1099), Higher: 856, Lower: 1012, Time: 2.90372s\n",
      "Epoch 112: Accuracy: 18.65332% (205/1099), Higher: 843, Lower: 1016, Time: 2.97541s\n",
      "Epoch 113: Accuracy: 18.65332% (205/1099), Higher: 858, Lower: 1012, Time: 2.93009s\n",
      "Epoch 114: Accuracy: 18.65332% (205/1099), Higher: 858, Lower: 1012, Time: 2.93162s\n",
      "Epoch 115: Accuracy: 18.65332% (205/1099), Higher: 858, Lower: 1012, Time: 2.89386s\n",
      "Epoch 116: Accuracy: 18.56233% (204/1099), Higher: 856, Lower: 1014, Time: 2.94376s\n",
      "Epoch 117: Accuracy: 18.56233% (204/1099), Higher: 856, Lower: 1015, Time: 2.87783s\n",
      "Epoch 118: Accuracy: 18.56233% (204/1099), Higher: 856, Lower: 1017, Time: 2.93275s\n",
      "Epoch 119: Accuracy: 18.65332% (205/1099), Higher: 856, Lower: 1011, Time: 2.92492s\n",
      "Epoch 120: Accuracy: 18.56233% (204/1099), Higher: 861, Lower: 1013, Time: 2.92308s\n",
      "Epoch 121: Accuracy: 18.56233% (204/1099), Higher: 859, Lower: 1016, Time: 2.97538s\n",
      "Epoch 122: Accuracy: 18.56233% (204/1099), Higher: 859, Lower: 1015, Time: 2.95482s\n",
      "Epoch 123: Accuracy: 18.56233% (204/1099), Higher: 859, Lower: 1015, Time: 2.93650s\n",
      "Epoch 124: Accuracy: 18.56233% (204/1099), Higher: 859, Lower: 1015, Time: 2.91071s\n",
      "Epoch 125: Accuracy: 18.38035% (202/1099), Higher: 854, Lower: 1029, Time: 2.92891s\n",
      "Epoch 126: Accuracy: 18.47134% (203/1099), Higher: 854, Lower: 1025, Time: 2.89046s\n",
      "Epoch 127: Accuracy: 18.38035% (202/1099), Higher: 842, Lower: 1033, Time: 3.02971s\n",
      "Epoch 128: Accuracy: 18.38035% (202/1099), Higher: 846, Lower: 1029, Time: 3.05331s\n",
      "Epoch 129: Accuracy: 18.19836% (200/1099), Higher: 845, Lower: 1037, Time: 3.01308s\n",
      "Epoch 130: Accuracy: 18.19836% (200/1099), Higher: 846, Lower: 1034, Time: 2.91999s\n",
      "Epoch 131: Accuracy: 18.28935% (201/1099), Higher: 842, Lower: 1035, Time: 2.90287s\n",
      "Epoch 132: Accuracy: 18.28935% (201/1099), Higher: 848, Lower: 1034, Time: 2.91623s\n",
      "Epoch 133: Accuracy: 18.28935% (201/1099), Higher: 844, Lower: 1038, Time: 2.96572s\n",
      "Epoch 134: Accuracy: 18.28935% (201/1099), Higher: 846, Lower: 1035, Time: 3.00637s\n",
      "Epoch 135: Accuracy: 18.28935% (201/1099), Higher: 844, Lower: 1040, Time: 3.01895s\n",
      "Epoch 136: Accuracy: 18.28935% (201/1099), Higher: 841, Lower: 1036, Time: 2.98981s\n",
      "Epoch 137: Accuracy: 18.38035% (202/1099), Higher: 841, Lower: 1037, Time: 2.93206s\n",
      "Epoch 138: Accuracy: 18.28935% (201/1099), Higher: 844, Lower: 1036, Time: 2.95031s\n",
      "Epoch 139: Accuracy: 18.38035% (202/1099), Higher: 839, Lower: 1039, Time: 2.90735s\n",
      "Epoch 140: Accuracy: 18.38035% (202/1099), Higher: 840, Lower: 1039, Time: 2.93273s\n",
      "Epoch 141: Accuracy: 18.28935% (201/1099), Higher: 843, Lower: 1038, Time: 3.02996s\n",
      "Epoch 142: Accuracy: 18.28935% (201/1099), Higher: 838, Lower: 1042, Time: 2.97398s\n",
      "Epoch 143: Accuracy: 18.38035% (202/1099), Higher: 846, Lower: 1036, Time: 2.96571s\n",
      "Epoch 144: Accuracy: 18.65332% (205/1099), Higher: 847, Lower: 1034, Time: 2.98266s\n",
      "Epoch 145: Accuracy: 18.65332% (205/1099), Higher: 843, Lower: 1037, Time: 2.98913s\n",
      "Epoch 146: Accuracy: 18.74431% (206/1099), Higher: 845, Lower: 1034, Time: 2.99558s\n",
      "Epoch 147: Accuracy: 18.65332% (205/1099), Higher: 845, Lower: 1037, Time: 2.94255s\n",
      "Epoch 148: Accuracy: 18.65332% (205/1099), Higher: 847, Lower: 1036, Time: 2.96846s\n",
      "Epoch 149: Accuracy: 18.65332% (205/1099), Higher: 841, Lower: 1039, Time: 2.90953s\n",
      "Epoch 150: Accuracy: 18.65332% (205/1099), Higher: 847, Lower: 1032, Time: 2.89989s\n",
      "Epoch 151: Accuracy: 18.56233% (204/1099), Higher: 849, Lower: 1034, Time: 2.90682s\n",
      "Epoch 152: Accuracy: 18.65332% (205/1099), Higher: 844, Lower: 1035, Time: 2.89011s\n",
      "Epoch 153: Accuracy: 18.56233% (204/1099), Higher: 841, Lower: 1036, Time: 2.96570s\n",
      "Epoch 154: Accuracy: 18.65332% (205/1099), Higher: 839, Lower: 1038, Time: 3.03370s\n",
      "Epoch 155: Accuracy: 18.65332% (205/1099), Higher: 839, Lower: 1038, Time: 3.01002s\n",
      "Epoch 156: Accuracy: 18.74431% (206/1099), Higher: 834, Lower: 1038, Time: 2.89930s\n",
      "Epoch 157: Accuracy: 18.65332% (205/1099), Higher: 838, Lower: 1040, Time: 2.90362s\n",
      "Epoch 158: Accuracy: 18.56233% (204/1099), Higher: 841, Lower: 1044, Time: 2.88953s\n",
      "Epoch 159: Accuracy: 18.65332% (205/1099), Higher: 839, Lower: 1044, Time: 2.93618s\n",
      "Epoch 160: Accuracy: 18.56233% (204/1099), Higher: 841, Lower: 1044, Time: 3.02677s\n",
      "Epoch 161: Accuracy: 18.56233% (204/1099), Higher: 841, Lower: 1044, Time: 3.06554s\n",
      "Epoch 162: Accuracy: 18.56233% (204/1099), Higher: 838, Lower: 1044, Time: 2.98928s\n",
      "Epoch 163: Accuracy: 18.65332% (205/1099), Higher: 842, Lower: 1042, Time: 2.90560s\n",
      "Epoch 164: Accuracy: 18.56233% (204/1099), Higher: 842, Lower: 1043, Time: 2.90701s\n",
      "Epoch 165: Accuracy: 18.56233% (204/1099), Higher: 841, Lower: 1044, Time: 2.90386s\n",
      "Epoch 166: Accuracy: 18.47134% (203/1099), Higher: 840, Lower: 1050, Time: 2.90860s\n",
      "Epoch 167: Accuracy: 18.56233% (204/1099), Higher: 838, Lower: 1051, Time: 2.90939s\n",
      "Epoch 168: Accuracy: 18.56233% (204/1099), Higher: 838, Lower: 1052, Time: 2.95327s\n",
      "Epoch 169: Accuracy: 18.56233% (204/1099), Higher: 837, Lower: 1050, Time: 2.95422s\n",
      "Epoch 170: Accuracy: 18.56233% (204/1099), Higher: 837, Lower: 1050, Time: 2.90367s\n",
      "Epoch 171: Accuracy: 18.56233% (204/1099), Higher: 837, Lower: 1050, Time: 2.90302s\n",
      "Epoch 172: Accuracy: 18.56233% (204/1099), Higher: 837, Lower: 1049, Time: 2.87859s\n",
      "Epoch 173: Accuracy: 18.56233% (204/1099), Higher: 837, Lower: 1050, Time: 3.04093s\n",
      "Epoch 174: Accuracy: 18.56233% (204/1099), Higher: 835, Lower: 1052, Time: 3.04954s\n",
      "Epoch 175: Accuracy: 18.56233% (204/1099), Higher: 835, Lower: 1052, Time: 3.01752s\n",
      "Epoch 176: Accuracy: 18.56233% (204/1099), Higher: 836, Lower: 1052, Time: 2.93062s\n",
      "Epoch 177: Accuracy: 18.56233% (204/1099), Higher: 835, Lower: 1052, Time: 2.96776s\n",
      "Epoch 178: Accuracy: 18.56233% (204/1099), Higher: 835, Lower: 1052, Time: 2.94729s\n",
      "Epoch 179: Accuracy: 18.56233% (204/1099), Higher: 835, Lower: 1052, Time: 3.03351s\n",
      "Epoch 180: Accuracy: 18.56233% (204/1099), Higher: 835, Lower: 1052, Time: 3.04599s\n",
      "Epoch 181: Accuracy: 18.56233% (204/1099), Higher: 839, Lower: 1052, Time: 2.94049s\n",
      "Epoch 182: Accuracy: 18.56233% (204/1099), Higher: 839, Lower: 1052, Time: 2.98789s\n",
      "Epoch 183: Accuracy: 18.56233% (204/1099), Higher: 839, Lower: 1052, Time: 3.07419s\n",
      "Epoch 184: Accuracy: 18.56233% (204/1099), Higher: 837, Lower: 1052, Time: 2.98865s\n",
      "Epoch 185: Accuracy: 18.47134% (203/1099), Higher: 837, Lower: 1054, Time: 2.92846s\n",
      "Epoch 186: Accuracy: 18.47134% (203/1099), Higher: 836, Lower: 1061, Time: 3.06868s\n",
      "Epoch 187: Accuracy: 18.38035% (202/1099), Higher: 843, Lower: 1057, Time: 3.01724s\n",
      "Epoch 188: Accuracy: 18.38035% (202/1099), Higher: 843, Lower: 1057, Time: 2.97424s\n",
      "Epoch 189: Accuracy: 18.47134% (203/1099), Higher: 834, Lower: 1057, Time: 3.06101s\n",
      "Epoch 190: Accuracy: 18.38035% (202/1099), Higher: 847, Lower: 1057, Time: 2.90654s\n",
      "Epoch 191: Accuracy: 18.28935% (201/1099), Higher: 847, Lower: 1061, Time: 2.98323s\n",
      "Epoch 192: Accuracy: 18.19836% (200/1099), Higher: 848, Lower: 1061, Time: 2.94583s\n",
      "Epoch 193: Accuracy: 18.28935% (201/1099), Higher: 847, Lower: 1061, Time: 2.96193s\n",
      "Epoch 194: Accuracy: 18.28935% (201/1099), Higher: 847, Lower: 1061, Time: 2.90219s\n",
      "Epoch 195: Accuracy: 18.19836% (200/1099), Higher: 848, Lower: 1065, Time: 2.92041s\n",
      "Epoch 196: Accuracy: 18.19836% (200/1099), Higher: 849, Lower: 1060, Time: 2.94637s\n",
      "Epoch 197: Accuracy: 18.19836% (200/1099), Higher: 849, Lower: 1061, Time: 2.91751s\n",
      "Epoch 198: Accuracy: 18.10737% (199/1099), Higher: 851, Lower: 1060, Time: 2.89460s\n",
      "Epoch 199: Accuracy: 18.19836% (200/1099), Higher: 850, Lower: 1064, Time: 2.92648s\n"
     ]
    }
   ],
   "execution_count": 23
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:55:52.736684Z",
     "start_time": "2024-05-28T09:55:52.598533Z"
    }
   },
   "cell_type": "code",
   "source": [
    "plt.plot(accuracy2)\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.show()"
   ],
   "id": "d1c69e46186b7f1c",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 24
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Validate",
   "id": "b5f499e13f49da23"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:55:52.741147Z",
     "start_time": "2024-05-28T09:55:52.737505Z"
    }
   },
   "cell_type": "code",
   "source": [
    "val_set2 = data2.iloc[:, :-2]\n",
    "val_dose2 = data2.iloc[:, -1:].values"
   ],
   "id": "888fd9f9f365c8b8",
   "outputs": [],
   "execution_count": 25
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:55:55.769343Z",
     "start_time": "2024-05-28T09:55:52.742028Z"
    }
   },
   "cell_type": "code",
   "source": [
    "arms = np.arange(12)\n",
    "correct_count = 0\n",
    "count = 0\n",
    "heatmap2 = np.zeros((12, 12))\n",
    "\n",
    "for i in range(len(data2)):\n",
    "    count += 1\n",
    "    feature = np.array(val_set2.iloc[i], dtype=float)\n",
    "    # iterate through actions\n",
    "    p = {}\n",
    "    max_value = -np.inf\n",
    "    max_arm = 0\n",
    "    for arm in arms:\n",
    "        theta = np.matmul(np.linalg.inv(A_map2[arm]), b_map2[arm])\n",
    "        p = np.matmul(theta.T, feature) + \\\n",
    "            np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map2[arm])), feature))\n",
    "        if p == max_value:\n",
    "            if np.random.choice(a=[False, True]): max_arm = arm\n",
    "        elif p > max_value:\n",
    "            max_value = p\n",
    "            max_arm = arm\n",
    "\n",
    "    if max_arm == val_dose2[i][0]:\n",
    "        correct_count += 1\n",
    "    heatmap2[max_arm][int(val_dose2[i][0])] += 1\n",
    "\n",
    "accu = correct_count / count\n",
    "print(\"Validate: Accuracy: %.5f%% (%d/%d)\" % (accu * 100, correct_count, count))"
   ],
   "id": "839d99f465512920",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validate: Accuracy: 25.44363% (932/3663)\n"
     ]
    }
   ],
   "execution_count": 26
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:55:56.198665Z",
     "start_time": "2024-05-28T09:55:55.770381Z"
    }
   },
   "cell_type": "code",
   "source": [
    "ax = sns.heatmap(heatmap2, linewidth=0.5, annot=True, cmap='Blues', fmt='.20g')\n",
    "plt.show()"
   ],
   "id": "49a7c7410c44a2c3",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 27
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": [
    "## Long range\n",
    "### Filter"
   ],
   "id": "51c0c79a8d78922e"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:55:56.204776Z",
     "start_time": "2024-05-28T09:55:56.199620Z"
    }
   },
   "cell_type": "code",
   "source": [
    "data3 = data2.copy()\n",
    "data3.loc[data3['dose'] < 21,'type'] = 0\n",
    "data3.loc[(data3['dose'] >= 21) & (data3['dose'] < 49),'type'] = 1\n",
    "data3.loc[data3['dose'] >= 49,'type'] = 2\n"
   ],
   "id": "53e1ec4801a6ceb6",
   "outputs": [],
   "execution_count": 28
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Split",
   "id": "527abe7398f3f275"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:55:56.264875Z",
     "start_time": "2024-05-28T09:55:56.207474Z"
    }
   },
   "cell_type": "code",
   "source": [
    "data_size3 = len(data3)\n",
    "ratio3 = 0.7\n",
    "pointer = int(ratio3 * data_size3)\n",
    "train_set3 = data3.iloc[:pointer, :-2]\n",
    "test_set3 = data3.iloc[pointer:, :-2]\n",
    "train_dose3 = data3.iloc[:pointer, -1:].values\n",
    "test_dose3 = data3.iloc[pointer:, -1:].values"
   ],
   "id": "7775ff62f1d2fa02",
   "outputs": [],
   "execution_count": 29
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Train",
   "id": "6651f44cdb1f923d"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:59:21.108566Z",
     "start_time": "2024-05-28T09:55:56.265880Z"
    }
   },
   "cell_type": "code",
   "source": [
    "arms = np.arange(3)\n",
    "accuracy3 = []\n",
    "A_map3 = {}\n",
    "b_map3 = {}\n",
    "for arm in arms:\n",
    "    A_map3[arm] = np.identity(len(list(train_set3)))\n",
    "    b_map3[arm] = np.zeros(len(list(train_set3)))\n",
    "\n",
    "for epoch in range(200):\n",
    "    start_time = time.time()\n",
    "    # train\n",
    "    indices = np.random.permutation(list(range(len(train_set3))))\n",
    "    for i in indices:\n",
    "        feature = np.array(train_set3.iloc[i], dtype=float)\n",
    "        # iterate through actions\n",
    "        p = {}\n",
    "        max_value = -np.inf\n",
    "        max_arm = 0\n",
    "        for arm in arms:\n",
    "            theta = np.matmul(np.linalg.inv(A_map3[arm]), b_map3[arm])\n",
    "            p = np.matmul(theta.T, feature) + \\\n",
    "                np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map3[arm])), feature))\n",
    "\n",
    "            if p == max_value:\n",
    "                if np.random.choice(a=[False, True]): max_arm = arm\n",
    "            elif p > max_value:\n",
    "                max_value = p\n",
    "                max_arm = arm\n",
    "\n",
    "        # observe reward and update A and b\n",
    "        r = -abs(train_dose3[i][0]-max_arm)\n",
    "        A_map3[max_arm] += np.outer(feature, feature)\n",
    "        b_map3[max_arm] += r * feature\n",
    "\n",
    "    # test\n",
    "    correct_count = 0\n",
    "    count = 0\n",
    "    higher = 0\n",
    "    lower = 0\n",
    "    for i in range(len(test_set3)):\n",
    "        count += 1\n",
    "        feature = np.array(test_set3.iloc[i], dtype=float)\n",
    "        # iterate through actions\n",
    "        p = {}\n",
    "        max_value = -np.inf\n",
    "        max_arm = 0\n",
    "        for arm in arms:\n",
    "            theta = np.matmul(np.linalg.inv(A_map3[arm]), b_map3[arm])\n",
    "            p = np.matmul(theta.T, feature) + \\\n",
    "                np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map3[arm])), feature))\n",
    "            if p == max_value:\n",
    "                if np.random.choice(a=[False, True]): max_arm = arm\n",
    "            elif p > max_value:\n",
    "                max_value = p\n",
    "                max_arm = arm\n",
    "\n",
    "        if max_arm == test_dose3[i]:\n",
    "            correct_count += 1\n",
    "        elif max_arm < test_dose3[i]:\n",
    "            lower += test_dose3[i] - max_arm\n",
    "        else:\n",
    "            higher += max_arm - test_dose3[i]\n",
    "\n",
    "    end_time = time.time()\n",
    "    accu = correct_count / count\n",
    "    print(\"Epoch %d: Accuracy: %.5f%% (%d/%d), Higher: %d, Lower: %d, Time: %.5fs\" % (\n",
    "        epoch, accu * 100, correct_count, count, higher, lower, end_time - start_time))\n",
    "    accuracy3.append(accu)"
   ],
   "id": "c2cde1e167281f7f",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0: Accuracy: 63.14832% (694/1099), Higher: 241, Lower: 172, Time: 1.02400s\n",
      "Epoch 1: Accuracy: 66.15105% (727/1099), Higher: 202, Lower: 175, Time: 1.02533s\n",
      "Epoch 2: Accuracy: 66.51501% (731/1099), Higher: 194, Lower: 178, Time: 1.04251s\n",
      "Epoch 3: Accuracy: 66.42402% (730/1099), Higher: 187, Lower: 186, Time: 1.00770s\n",
      "Epoch 4: Accuracy: 66.24204% (728/1099), Higher: 186, Lower: 190, Time: 0.99082s\n",
      "Epoch 5: Accuracy: 66.51501% (731/1099), Higher: 186, Lower: 186, Time: 0.99930s\n",
      "Epoch 6: Accuracy: 66.24204% (728/1099), Higher: 184, Lower: 191, Time: 1.03060s\n",
      "Epoch 7: Accuracy: 65.96906% (725/1099), Higher: 190, Lower: 189, Time: 1.02595s\n",
      "Epoch 8: Accuracy: 66.15105% (727/1099), Higher: 186, Lower: 190, Time: 1.01796s\n",
      "Epoch 9: Accuracy: 66.15105% (727/1099), Higher: 189, Lower: 188, Time: 1.04962s\n",
      "Epoch 10: Accuracy: 65.87807% (724/1099), Higher: 189, Lower: 191, Time: 1.01472s\n",
      "Epoch 11: Accuracy: 65.96906% (725/1099), Higher: 190, Lower: 189, Time: 1.00227s\n",
      "Epoch 12: Accuracy: 65.78708% (723/1099), Higher: 189, Lower: 192, Time: 1.04440s\n",
      "Epoch 13: Accuracy: 66.06005% (726/1099), Higher: 188, Lower: 190, Time: 1.07230s\n",
      "Epoch 14: Accuracy: 66.15105% (727/1099), Higher: 190, Lower: 187, Time: 1.06189s\n",
      "Epoch 15: Accuracy: 66.06005% (726/1099), Higher: 190, Lower: 188, Time: 1.06381s\n",
      "Epoch 16: Accuracy: 66.06005% (726/1099), Higher: 190, Lower: 188, Time: 1.04853s\n",
      "Epoch 17: Accuracy: 66.15105% (727/1099), Higher: 189, Lower: 188, Time: 1.04468s\n",
      "Epoch 18: Accuracy: 65.87807% (724/1099), Higher: 192, Lower: 188, Time: 1.00311s\n",
      "Epoch 19: Accuracy: 65.87807% (724/1099), Higher: 191, Lower: 189, Time: 1.00641s\n",
      "Epoch 20: Accuracy: 65.78708% (723/1099), Higher: 190, Lower: 191, Time: 1.03570s\n",
      "Epoch 21: Accuracy: 65.78708% (723/1099), Higher: 190, Lower: 191, Time: 1.04621s\n",
      "Epoch 22: Accuracy: 65.78708% (723/1099), Higher: 191, Lower: 190, Time: 0.99543s\n",
      "Epoch 23: Accuracy: 65.78708% (723/1099), Higher: 191, Lower: 190, Time: 0.99818s\n",
      "Epoch 24: Accuracy: 65.78708% (723/1099), Higher: 190, Lower: 191, Time: 1.00735s\n",
      "Epoch 25: Accuracy: 65.69609% (722/1099), Higher: 191, Lower: 192, Time: 1.01050s\n",
      "Epoch 26: Accuracy: 65.78708% (723/1099), Higher: 190, Lower: 192, Time: 1.00614s\n",
      "Epoch 27: Accuracy: 65.96906% (725/1099), Higher: 188, Lower: 192, Time: 1.01331s\n",
      "Epoch 28: Accuracy: 65.87807% (724/1099), Higher: 191, Lower: 190, Time: 1.01434s\n",
      "Epoch 29: Accuracy: 65.87807% (724/1099), Higher: 191, Lower: 190, Time: 1.01260s\n",
      "Epoch 30: Accuracy: 65.87807% (724/1099), Higher: 191, Lower: 190, Time: 1.01223s\n",
      "Epoch 31: Accuracy: 65.78708% (723/1099), Higher: 191, Lower: 191, Time: 1.02495s\n",
      "Epoch 32: Accuracy: 66.06005% (726/1099), Higher: 189, Lower: 190, Time: 1.01539s\n",
      "Epoch 33: Accuracy: 65.87807% (724/1099), Higher: 191, Lower: 190, Time: 1.01929s\n",
      "Epoch 34: Accuracy: 66.06005% (726/1099), Higher: 191, Lower: 188, Time: 1.00696s\n",
      "Epoch 35: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.00913s\n",
      "Epoch 36: Accuracy: 65.87807% (724/1099), Higher: 191, Lower: 190, Time: 1.01355s\n",
      "Epoch 37: Accuracy: 65.87807% (724/1099), Higher: 191, Lower: 190, Time: 1.04657s\n",
      "Epoch 38: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.02605s\n",
      "Epoch 39: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.04037s\n",
      "Epoch 40: Accuracy: 65.78708% (723/1099), Higher: 191, Lower: 191, Time: 1.00889s\n",
      "Epoch 41: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.00901s\n",
      "Epoch 42: Accuracy: 66.15105% (727/1099), Higher: 190, Lower: 188, Time: 1.00598s\n",
      "Epoch 43: Accuracy: 66.15105% (727/1099), Higher: 190, Lower: 188, Time: 1.01129s\n",
      "Epoch 44: Accuracy: 66.15105% (727/1099), Higher: 190, Lower: 188, Time: 0.99979s\n",
      "Epoch 45: Accuracy: 66.15105% (727/1099), Higher: 190, Lower: 188, Time: 1.01033s\n",
      "Epoch 46: Accuracy: 66.06005% (726/1099), Higher: 190, Lower: 189, Time: 1.00636s\n",
      "Epoch 47: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.01459s\n",
      "Epoch 48: Accuracy: 66.06005% (726/1099), Higher: 190, Lower: 189, Time: 1.01066s\n",
      "Epoch 49: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.02893s\n",
      "Epoch 50: Accuracy: 65.87807% (724/1099), Higher: 191, Lower: 190, Time: 0.99636s\n",
      "Epoch 51: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.03910s\n",
      "Epoch 52: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.01054s\n",
      "Epoch 53: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 0.99286s\n",
      "Epoch 54: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.02641s\n",
      "Epoch 55: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.02123s\n",
      "Epoch 56: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.01291s\n",
      "Epoch 57: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.01400s\n",
      "Epoch 58: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.01522s\n",
      "Epoch 59: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.01930s\n",
      "Epoch 60: Accuracy: 66.06005% (726/1099), Higher: 191, Lower: 188, Time: 1.02066s\n",
      "Epoch 61: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.02403s\n",
      "Epoch 62: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.01127s\n",
      "Epoch 63: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.04579s\n",
      "Epoch 64: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.02232s\n",
      "Epoch 65: Accuracy: 65.96906% (725/1099), Higher: 191, Lower: 189, Time: 1.03120s\n",
      "Epoch 66: Accuracy: 66.06005% (726/1099), Higher: 190, Lower: 189, Time: 1.05104s\n",
      "Epoch 67: Accuracy: 66.06005% (726/1099), Higher: 190, Lower: 189, Time: 1.02423s\n",
      "Epoch 68: Accuracy: 65.96906% (725/1099), Higher: 190, Lower: 190, Time: 1.04586s\n",
      "Epoch 69: Accuracy: 66.06005% (726/1099), Higher: 190, Lower: 189, Time: 1.03346s\n",
      "Epoch 70: Accuracy: 66.06005% (726/1099), Higher: 190, Lower: 189, Time: 1.02661s\n",
      "Epoch 71: Accuracy: 66.06005% (726/1099), Higher: 190, Lower: 189, Time: 1.04706s\n",
      "Epoch 72: Accuracy: 66.06005% (726/1099), Higher: 190, Lower: 189, Time: 0.99933s\n",
      "Epoch 73: Accuracy: 66.15105% (727/1099), Higher: 190, Lower: 188, Time: 1.00262s\n",
      "Epoch 74: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.06671s\n",
      "Epoch 75: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.07432s\n",
      "Epoch 76: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.08154s\n",
      "Epoch 77: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.03833s\n",
      "Epoch 78: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 0.99277s\n",
      "Epoch 79: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 0.99592s\n",
      "Epoch 80: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 0.99760s\n",
      "Epoch 81: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 0.99728s\n",
      "Epoch 82: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 0.99853s\n",
      "Epoch 83: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00588s\n",
      "Epoch 84: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00630s\n",
      "Epoch 85: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00875s\n",
      "Epoch 86: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99836s\n",
      "Epoch 87: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00526s\n",
      "Epoch 88: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00377s\n",
      "Epoch 89: Accuracy: 66.51501% (731/1099), Higher: 187, Lower: 187, Time: 0.99920s\n",
      "Epoch 90: Accuracy: 66.51501% (731/1099), Higher: 187, Lower: 187, Time: 1.02237s\n",
      "Epoch 91: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00554s\n",
      "Epoch 92: Accuracy: 66.51501% (731/1099), Higher: 187, Lower: 187, Time: 1.00055s\n",
      "Epoch 93: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99796s\n",
      "Epoch 94: Accuracy: 66.42402% (730/1099), Higher: 187, Lower: 188, Time: 0.99603s\n",
      "Epoch 95: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00168s\n",
      "Epoch 96: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.00509s\n",
      "Epoch 97: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.00507s\n",
      "Epoch 98: Accuracy: 66.15105% (727/1099), Higher: 190, Lower: 188, Time: 1.03969s\n",
      "Epoch 99: Accuracy: 66.15105% (727/1099), Higher: 190, Lower: 188, Time: 1.02197s\n",
      "Epoch 100: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.02235s\n",
      "Epoch 101: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.01136s\n",
      "Epoch 102: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00662s\n",
      "Epoch 103: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.01684s\n",
      "Epoch 104: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.00775s\n",
      "Epoch 105: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.00339s\n",
      "Epoch 106: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.00651s\n",
      "Epoch 107: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.00435s\n",
      "Epoch 108: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.00597s\n",
      "Epoch 109: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.01769s\n",
      "Epoch 110: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.01585s\n",
      "Epoch 111: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.01331s\n",
      "Epoch 112: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00725s\n",
      "Epoch 113: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.01130s\n",
      "Epoch 114: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.04950s\n",
      "Epoch 115: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.05715s\n",
      "Epoch 116: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.06591s\n",
      "Epoch 117: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.04142s\n",
      "Epoch 118: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00843s\n",
      "Epoch 119: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.04010s\n",
      "Epoch 120: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.06683s\n",
      "Epoch 121: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00629s\n",
      "Epoch 122: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00045s\n",
      "Epoch 123: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.01112s\n",
      "Epoch 124: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00729s\n",
      "Epoch 125: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00968s\n",
      "Epoch 126: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00854s\n",
      "Epoch 127: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.00676s\n",
      "Epoch 128: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.01917s\n",
      "Epoch 129: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.02135s\n",
      "Epoch 130: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.02233s\n",
      "Epoch 131: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.03093s\n",
      "Epoch 132: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.03846s\n",
      "Epoch 133: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.09042s\n",
      "Epoch 134: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.02065s\n",
      "Epoch 135: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.02531s\n",
      "Epoch 136: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.02434s\n",
      "Epoch 137: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.02013s\n",
      "Epoch 138: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.03368s\n",
      "Epoch 139: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.05970s\n",
      "Epoch 140: Accuracy: 66.33303% (729/1099), Higher: 188, Lower: 188, Time: 1.01000s\n",
      "Epoch 141: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.08619s\n",
      "Epoch 142: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.04423s\n",
      "Epoch 143: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.01983s\n",
      "Epoch 144: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.02234s\n",
      "Epoch 145: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.05544s\n",
      "Epoch 146: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.10221s\n",
      "Epoch 147: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.05267s\n",
      "Epoch 148: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.07272s\n",
      "Epoch 149: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.02161s\n",
      "Epoch 150: Accuracy: 66.24204% (728/1099), Higher: 189, Lower: 188, Time: 1.04296s\n",
      "Epoch 151: Accuracy: 66.33303% (729/1099), Higher: 189, Lower: 187, Time: 1.02870s\n",
      "Epoch 152: Accuracy: 66.33303% (729/1099), Higher: 189, Lower: 187, Time: 1.03031s\n",
      "Epoch 153: Accuracy: 66.33303% (729/1099), Higher: 189, Lower: 187, Time: 1.04454s\n",
      "Epoch 154: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.09779s\n",
      "Epoch 155: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.08060s\n",
      "Epoch 156: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.08038s\n",
      "Epoch 157: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.03477s\n",
      "Epoch 158: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.03076s\n",
      "Epoch 159: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.05791s\n",
      "Epoch 160: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00168s\n",
      "Epoch 161: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.03749s\n",
      "Epoch 162: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.03416s\n",
      "Epoch 163: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00300s\n",
      "Epoch 164: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.01666s\n",
      "Epoch 165: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.05321s\n",
      "Epoch 166: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.03783s\n",
      "Epoch 167: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00095s\n",
      "Epoch 168: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00323s\n",
      "Epoch 169: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.04618s\n",
      "Epoch 170: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.02545s\n",
      "Epoch 171: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.01344s\n",
      "Epoch 172: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.02573s\n",
      "Epoch 173: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.06851s\n",
      "Epoch 174: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.05145s\n",
      "Epoch 175: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.02779s\n",
      "Epoch 176: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.03167s\n",
      "Epoch 177: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.05777s\n",
      "Epoch 178: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.08010s\n",
      "Epoch 179: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.06344s\n",
      "Epoch 180: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.01928s\n",
      "Epoch 181: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.06662s\n",
      "Epoch 182: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.05329s\n",
      "Epoch 183: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.03594s\n",
      "Epoch 184: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99855s\n",
      "Epoch 185: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00128s\n",
      "Epoch 186: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99581s\n",
      "Epoch 187: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99516s\n",
      "Epoch 188: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99929s\n",
      "Epoch 189: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99440s\n",
      "Epoch 190: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00659s\n",
      "Epoch 191: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99569s\n",
      "Epoch 192: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99540s\n",
      "Epoch 193: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99964s\n",
      "Epoch 194: Accuracy: 66.51501% (731/1099), Higher: 188, Lower: 186, Time: 1.00024s\n",
      "Epoch 195: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99222s\n",
      "Epoch 196: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 0.99758s\n",
      "Epoch 197: Accuracy: 66.51501% (731/1099), Higher: 188, Lower: 186, Time: 0.99923s\n",
      "Epoch 198: Accuracy: 66.42402% (730/1099), Higher: 188, Lower: 187, Time: 1.00773s\n",
      "Epoch 199: Accuracy: 66.60601% (732/1099), Higher: 188, Lower: 185, Time: 1.00488s\n"
     ]
    }
   ],
   "execution_count": 30
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:59:21.251154Z",
     "start_time": "2024-05-28T09:59:21.109455Z"
    }
   },
   "cell_type": "code",
   "source": [
    "plt.plot(accuracy3)\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.show()"
   ],
   "id": "bb4d8376aba6f912",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 31
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Validate",
   "id": "ccbd96449e3a1325"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:59:21.255957Z",
     "start_time": "2024-05-28T09:59:21.252309Z"
    }
   },
   "cell_type": "code",
   "source": [
    "val_set3 = data3.iloc[:, :-2]\n",
    "val_dose3 = data3.iloc[:, -1:].values"
   ],
   "id": "79b356809f2bd8e4",
   "outputs": [],
   "execution_count": 32
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:59:22.313192Z",
     "start_time": "2024-05-28T09:59:21.257095Z"
    }
   },
   "cell_type": "code",
   "source": [
    "arms = np.arange(3)\n",
    "correct_count = 0\n",
    "count = 0\n",
    "heatmap3 = np.zeros((3, 3))\n",
    "\n",
    "for i in range(len(data3)):\n",
    "    count += 1\n",
    "    feature = np.array(val_set3.iloc[i], dtype=float)\n",
    "    # iterate through actions\n",
    "    p = {}\n",
    "    max_value = -np.inf\n",
    "    max_arm = 0\n",
    "    for arm in arms:\n",
    "        theta = np.matmul(np.linalg.inv(A_map3[arm]), b_map3[arm])\n",
    "        p = np.matmul(theta.T, feature) + \\\n",
    "            np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map3[arm])), feature))\n",
    "        if p == max_value:\n",
    "            if np.random.choice(a=[False, True]): max_arm = arm\n",
    "        elif p > max_value:\n",
    "            max_value = p\n",
    "            max_arm = arm\n",
    "\n",
    "    if max_arm == val_dose3[i][0]:\n",
    "        correct_count += 1\n",
    "    heatmap3[max_arm][int(val_dose3[i][0])] += 1\n",
    "\n",
    "accu = correct_count / count\n",
    "print(\"Validate: Accuracy: %.5f%% (%d/%d)\" % (accu * 100, correct_count, count))"
   ],
   "id": "1b1a770a3bdff382",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validate: Accuracy: 71.19847% (2608/3663)\n"
     ]
    }
   ],
   "execution_count": 33
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T09:59:22.462277Z",
     "start_time": "2024-05-28T09:59:22.314443Z"
    }
   },
   "cell_type": "code",
   "source": [
    "ax = sns.heatmap(heatmap3, linewidth=0.5, annot=True, cmap='Blues', fmt='.20g')\n",
    "plt.show()"
   ],
   "id": "5685e253348e98cc",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 34
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T11:37:18.357747Z",
     "start_time": "2024-05-28T11:34:54.338165Z"
    }
   },
   "cell_type": "code",
   "source": [
    "arms = np.arange(12)\n",
    "loss = []\n",
    "A_map0 = {}\n",
    "b_map0 = {}\n",
    "for arm in arms:\n",
    "    A_map0[arm] = np.identity(len(list(val_set)))\n",
    "    b_map0[arm] = np.zeros(len(list(val_set)))\n",
    "\n",
    "for epoch in range(20):\n",
    "    # train\n",
    "    indices = np.random.permutation(list(range(len(val_set))))\n",
    "    for i in indices:\n",
    "        feature = np.array(val_set.iloc[i], dtype=float)\n",
    "        # iterate through actions\n",
    "        p = {}\n",
    "        max_value = -np.inf\n",
    "        max_arm = 0\n",
    "        for arm in arms:\n",
    "            theta = np.matmul(np.linalg.inv(A_map0[arm]), b_map0[arm])\n",
    "            p = np.matmul(theta.T, feature) + \\\n",
    "                np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map0[arm])), feature))\n",
    "\n",
    "            if p == max_value:\n",
    "                if np.random.choice(a=[False, True]): max_arm = arm\n",
    "            elif p > max_value:\n",
    "                max_value = p\n",
    "                max_arm = arm\n",
    "\n",
    "        # observe reward and update A and b\n",
    "        r = -abs(val_dose[i][0]-max_arm)\n",
    "        A_map0[max_arm] += np.outer(feature, feature)\n",
    "        b_map0[max_arm] += r * feature\n",
    "        \n",
    "        if len(loss) == 0:\n",
    "            loss.append(-r)\n",
    "        else:\n",
    "            loss.append((loss[-1]*len(loss)-r)/(len(loss)+1))\n",
    "            \n",
    "plt.plot(loss)\n",
    "plt.xlabel('Learning time')\n",
    "plt.ylabel('Average Loss')\n",
    "plt.show()"
   ],
   "id": "4e858a516725562d",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ],
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAioAAAGwCAYAAACHJU4LAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAAA6WUlEQVR4nO3deXwU9f3H8ffskUC4r4IigiAJAsFEVCrEIBRpVVBUfvWo6I96ofbRKoIH9WcrgnhiK1pKtVVbFKWoiMLvJx71AuVQkFPumyDIEQIJSXb2+/tjs0tCEsjixhl2Xs/HgwfZmdmZz2eXkHe+850dyxhjBAAA4EI+pwsAAACoDkEFAAC4FkEFAAC4FkEFAAC4FkEFAAC4FkEFAAC4FkEFAAC4VsDpAn6IcDisUCgkn88ny7KcLgcAANSAMUbhcFiBQEA+39HHTE7ooBIKhbR06VKnywAAAMchMzNTKSkpR93mhA4q0RSWmZkpv9+f0H3btq2lS5fWyr7dyos9S97s24s9S97s24s9S/Tt9r6jdR5rNEU6wYNK9HSP3++vtTekNvftVl7sWfJm317sWfJm317sWaJvt6vJtA0m0wIAANciqAAAANciqAAAANciqAAAANciqAAAANciqAAAANciqAAAANciqAAAANciqAAAANciqAAAANdy9CP0+/btq23btlVafu211+oPf/iDAxUBAAA3cTSoTJs2TbZtxx6vWbNGQ4cO1S9+8QsHqwIAAG7haFBp2rRphcd/+9vfdOqpp+rcc891qKLDikrsY28EAABqlWvunlxSUqIZM2Zo6NChNbqbYnnlR2USYfV3BbrsL1/o4g51lZnpncASfR0T/Xq6nRf79mLPkjf79mLPEn27ve946rOMMaYWa6mxWbNmacSIEfrPf/6jli1b1ug5tm1r8eLFCa/l881FenpevjJ/kqI/9m567CcAAIC4ZWVlye/3H3Ub14yovPHGG8rNza1xSCkvMzPzmI3GY4svT5r3Ta3s281s29bSpUs91bPkzb692LPkzb692LNE327vO1pnTbgiqGzbtk1z587VhAkTjuv5fr8/oW+Iz3f4qu1E7/tE4MWeJW/27cWeJW/27cWeJfpOBq74HJU333xTzZo10wUXXOB0KQAAwEUcDyrhcFhvvvmmBg0apEDAFQM8AADAJRwPKnPnztX27dt15ZVXOl0KAABwGceHMHJycrRq1SqnywAAAC7k+IgKAABAdQgqAADAtQgqAADAtQgqAADAtQgqAADAtQgqR+GOuyABAOBdBBUAAOBaBJUqWE4XAAAAJBFUAACAixFUAACAaxFUAACAaxFUAACAaxFUAACAaxFUAACAaxFUAACAaxFUAACAaxFUAACAaxFUAACAaxFUjoJ7EgIA4CyCCgAAcC2CCgAAcC2CShUsbp8MAIArEFQAAIBrEVQAAIBrEVQAAIBrEVQAAIBrEVQAAIBrEVQAAIBrEVQAAIBrEVQAAIBrEVQAAIBrEVSOitsSAgDgJIIKAABwLYIKAABwLYIKAABwLYJKFSxx+2QAANyAoAIAAFyLoAIAAFyLoAIAAFyLoAIAAFyLoAIAAFyLoAIAAFyLoAIAAFyLoAIAAFyLoHIUhnsSAgDgKIIKAABwLYIKAABwLYIKAABwLceDynfffacRI0aoR48eOvPMM3XZZZdp2bJlTpcFAABcIODkwfPz83XNNdeoR48eev7559W0aVNt2bJFDRs2dLIsWdw8GQAAV3A0qDz//PNq1aqVxo0bF1t2yimnxL0f27YTWZbC4XCt7dvNor16qWfJm317sWfJm317sWeJvt3edzz1WcY4dxHuxRdfrJycHO3YsUMLFixQy5Ytde211+qXv/xljZ5v27YWL16c8Lq+2HpIT36xT52bB/Vwn2YJ3z8AAJCysrLk9/uPuo2jIypbtmzRlClTNHToUA0bNkxLlizRmDFjlJKSokGDBtV4P5mZmcdsNB55gR3SF4trZd9uZtu2li5d6qmeJW/27cWeJW/27cWeJfp2e9/ROmvC0aBijFHXrl01fPhwSVLnzp21du1aTZkyJa6g4vf7E/qG+HyH5xgnet8nAi/2LHmzby/2LHmzby/2LNF3MnD0qp8WLVqoQ4cOFZa1b99e27dvd6giAADgJo4GlbPOOksbNmyosGzjxo1q3bq1QxUBAAA3cTSo3HDDDfrmm2/017/+VZs2bdI777yjqVOn6tprr3WyLAAA4BKOzlHp1q2bnn32WY0fP17PPfecTjnlFI0aNUqXXnqpk2XFcE9CAACc5WhQkaQ+ffqoT58+TpcBAABcyPGP0AcAAKgOQQUAALgWQQUAALgWQaUK3JMQAAB3IKgAAADXIqgAAADXIqgAAADXIqgAAADXIqgAAADXIqgAAADXIqgAAADXIqgAAADXIqgAAADXIqgAAADXIqgAAADXIqgAAADXIqgAAADXIqhUweL2yQAAuAJBBQAAuBZBBQAAuBZBBQAAuBZBBQAAuBZBBQAAuBZBBQAAuBZB5SiMcboCAAC8jaACAABci6ACAABci6ACAABci6ACAABci6ACAABci6ACAABci6BSJW6fDACAGxBUAACAaxFUAACAaxFUAACAaxFUAACAaxFUjoJb/QAA4CyCCgAAcC2CCgAAcC2CCgAAcC2CCgAAcC2CCgAAcC2CCgAAcC2CCgAAcC2CCgAAcC2CShUsbp4MAIArEFQAAIBrBZw8+IQJE/Tss89WWNa8eXPNmTPHoYoAAICbOBpUJKljx4568cUXY4/9fr+D1QAAADdxPKj4/X61aNHC6TIAAIALOR5UNm3apJycHKWkpOjMM8/U8OHD1aZNm7j2Ydt2QmsKh8O1tm83i/bqpZ4lb/btxZ4lb/btxZ4l+nZ73/HUZxljTC3WclSffPKJDh06pHbt2mn37t2aOHGi1q9fr3fffVdNmjQ55vNt29bixYsTXte8bYf0+Nx9ymgW1CN9myV8/wAAQMrKyjrmlA9HR1R69+5d4XFWVpYuvPBCTZ8+XUOHDq3xfjIzMxM6t2VnynfS3EW1sm83s21bS5cu9VTPkjf79mLPkjf79mLPEn27ve9onTXh+Kmf8tLS0pSenq6NGzfG9Ty/35/QN8TnO3zVdqL3fSLwYs+SN/v2Ys+SN/v2Ys8SfScDV32OSklJidatW8fkWgAAIMnhEZXHHntMffr00UknnaQ9e/Zo4sSJOnDggC6//HInywIAAC7haFDZsWOHhg8frn379qlJkybKysrS1KlT1bp1ayfLAgAALuFoUHn66aedPDwAAHC5hMxR2b9/fyJ2AwAAUEHcQeVvf/ubZs2aFXv8u9/9Tj169ND555+vb7/9NqHFOYWbJwMA4A5xB5XXX39drVq1kiTNmTNHc+fO1fPPP6/c3Fw9/vjjCS8QAAB4V9xzVHbt2qWTTjpJkvSf//xHF110kXJyctS6dWv98pe/THiBAADAu+IeUWnYsKHy8vIkSZ999pnOO+88SZIxxvX3FoiXczcXAAAA0nGMqPTv318jRoxQ27ZttW/fPuXm5kqSvv32W7Vt2zbhBQIAAO+KO6jcf//9at26tfLy8jRy5EjVq1dPkrRz505de+21CS8QAAB4V9xBJRgM6sYbb6y0/L//+78TUQ8AAEBM3HNU3nrrLX388cexx48//rjOPvtsXX311dq2bVsiawMAAB4Xd1D561//qtTUVEnSokWL9Morr2jkyJFq3Lixxo0bl/ACAQCAd8V96mfHjh2xSbMffPCBfv7zn+uqq67SWWedpSFDhiS8QAAA4F1xj6ikpaVp3759kiIf+NazZ09JUmpqqoqLixNaHAAA8La4R1R69uypBx54QGeccYY2btyo3r17S5LWrFnDXY8BAEBCxT2i8oc//EFZWVnas2ePnnnmGTVp0kSStHz5cl1yySUJLxAAAHhX3CMqDRs21IMPPlhp+W9/+9uEFOQGlsVtCQEAcIO4g4ok7d+/X9OmTdO6detkWZY6dOigwYMHq0GDBomuDwAAeFjcp36WLl2qCy+8UC+99JLy8/O1d+9evfTSS+rXr5+WL19eGzUCAACPintEZdy4cerbt68efvhhBQKRp4dCIT3wwAN65JFH9MorryS8SKdwT0IAAJwV94jKsmXLdNNNN8VCiiQFAgHddNNNWrZsWUKLAwAA3hZ3UKlfv77y8vIqLc/Ly4vdoBAAACAR4g4qF198sX7/+99r1qxZysvL044dOzRz5kw98MADXJ4MAAASKu45Kvfcc0/sb9u2IzsJBHTNNddoxIgRia0OAAB4WtxBJSUlRQ888IDuvvtubd68WcYYtW3bVoFAQLt27dLJJ59cG3UCAAAPivvUT1TdunWVkZGhTp06qW7dulq3bp1+9rOfJbI2AADgcccdVAAAAGobQQUAALgWQQUAALhWjSfTfvvtt0ddv379+h9cDAAAQHk1DiqDBg2SZVkypvIHy0eXJ8tdh5OjCwAATnw1DioffvhhbdYBAABQSY2DSuvWrWuzDgAAgEqYTAsAAFyLoAIAAFyLoAIAAFyLoAIAAFzruIJKKBTS3Llz9dprr+nAgQOSpO+++04HDx5MaHEAAMDb4r578rZt23TTTTcpLy9PJSUl6tWrl+rXr68XXnhBxcXFGj16dG3UCQAAPCjuEZWxY8eqa9eumj9/vlJTU2PLL7zwQn355ZcJLQ4AAHhb3CMqX331laZMmaKUlJQKy08++WR99913CSsMAAAg7hEVY4zC4XCl5Tt27FC9evUSUhQAAIB0HEGlZ8+eevnllyssO3jwoCZMmKDevXsnrDAAAIC4T/3cf//9uv7663XxxRerpKREI0aM0MaNG9WkSRONHz++NmoEAAAeFXdQadmypd5++229++67WrFihcLhsAYPHqyBAweqTp06tVHjjy5JbgINAMAJL+6gIkl16tTR4MGDE12L6xjjdAUAAHhb3EHlww8/rHK5ZVlKTU3VqaeeqjZt2vzgwgAAAOIOKnfccYcsy5I5YrghusyyLHXv3l3PPfecGjVqlLBCAQCA98R91c+LL76ozMxMvfjii1q4cKEWLlyoF198Ud26ddOkSZM0efJk7du3T4899lht1AsAADwk7hGVsWPHavTo0TrrrLNiy8477zylpKTowQcf1MyZMzVq1CiNGjUqoYUCAADviXtEZfPmzapfv36l5fXr19eWLVskSW3bttXevXvj2u+kSZOUkZGhsWPHxlsSAABIUnEHlS5duujxxx/Xnj17Ysv27NmjJ554QpmZmZKkTZs2qVWrVjXe55IlS/T6668rIyMj3nIAAEASO65TP7fffrtyc3N10kknybIsbd++XW3atNFf/vIXSVJhYaFuu+22Gu3v4MGDGjlypMaMGaOJEyfGW44kybbt43pe9fuL3CJg7d7ShO/bzaK9eqlnyZt9e7FnyZt9e7Fnib7d3nc89VnmyMt3asAYo88++0wbN26UMUbt27dXr1695PPFPUCje++9V40aNdKoUaM0ZMgQderUSb///e9r9FzbtrV48eK4j3ksk77K1+z1RZKkN/6r5iNDAACg5rKysuT3+4+6zXF94JtlWcrNzVVubu5xFRY1c+ZMrVixQtOmTftB+8nMzDxmo/HYt2CepKJa2beb2batpUuXeqpnyZt9e7FnyZt9e7Fnib7d3ne0zpo4rqBSWFioBQsWaPv27SotLa2w7vrrr6/RPvLy8jR27Fj94x//UGpq6vGUEeP3+xP8hhz+DP3E79v9vNiz5M2+vdiz5M2+vdizRN/JIO6gsmLFCt1yyy0qKipSUVGRGjVqpL1796pu3bpq2rRpjYPK8uXLtXv3bl1xxRWxZbZta8GCBXrllVe0dOnSpHmRAQDA8Yk7qIwbN059+vTRH//4R5199tmaOnWqAoGARo4cWeOQIkk//elP9c4771RYdv/996t9+/a6+eabCSkAACD+oLJy5Uo99NBDsWGlkpIStWnTRiNHjtS9996r/v3712g/9evXV3p6eoVlaWlpaty4caXlPzrungwAgCvEfZlOIBCQZUV+kjdr1kzbt2+XFAkeeXl5ia0OAAB4WtwjKp07d9ayZct02mmnqUePHnrmmWe0d+9evf322z94JORf//rXD3o+AABILnGPqNx1111q0aKFJOnOO+9U48aN9cc//lF79uzRww8/nPACAQCAd8U1omKMUdOmTdWxY0dJUtOmTfX888/XSmEAAABxjagYY/Tzn/9cO3bsqK16AAAAYuIKKj6fT23bttW+fftqqRyXiPumAgAAoDbEPUdl5MiRevzxx7V69eraqAcAACAm7qt+Ro4cqaKiIl122WUKBoOqU6dOhfXz589PWHEAAMDb4g4qo0aNqo06AAAAKok7qFx++eW1UQcAAEAlcc9RkaTNmzfr6aef1vDhw7V7925J0qeffqo1a9YktDinGGbTAgDgCnEHlfnz52vgwIFasmSJZs+ercLCQknSqlWrNGHChIQXCAAAvCvuoPLUU0/pzjvv1IsvvqhgMBhb3qNHDy1atCihxTnl5vPbO10CAADQcQSV1atXq1+/fpWWN23aNGk+X6Vts3qSpIapx3VmDAAAJEjcP4kbNGigXbt2VVq+cuVKtWzZMiFFOa3s5tCSYa4KAABOijuoDBgwQE8++aR27doly7IUDof11Vdf6bHHHtOgQYNqocQfn3XsTQAAwI/guO6efNJJJyk3N1eFhYW65JJLdN111yk7O1u33XZbbdToGMZTAABwVtyfoxIMBvXUU0/pd7/7nVasWKFwOKzOnTurXbt2tVCeM6KnfggqAAA4K+6gMn/+fJ177rk69dRTdeqpp9ZGTS5AUgEAwA3iDiq//vWv1bx5cw0YMECXXnqp0tPTa6MuR1lMUgEAwBXinqPy6aef6qabbtLChQt16aWXauDAgXr++ee1Y8eO2qjPUQyoAADgrLiDStOmTXXdddfptdde0/vvv6+LL75Yb7/9tvr27avrr7++Nmr80cWuTna0CgAAEPepn/LatGmjW265RZ06ddKf//xnLViwIFF1OcpiNi0AAK5w3EHlq6++0jvvvKP33ntPxcXF6tu3r+66665E1uYYRlQAAHCHuIPK+PHj9e6772rnzp3q2bOnRo0apX79+qlu3bq1UZ8jmEwLAIA7HNflyTfeeKMuuugiNW3atMK6lStX6owzzkhYcU5jRAUAAGfFHVRee+21Co8LCgo0Y8YMTZs2Td9++61WrlyZsOKcYvE5KgAAuMJxz1H54osv9MYbb+j999/XySefrP79+2vs2LGJrM0xzKUFAMAd4goqO3bs0Jtvvqk33nhDRUVFuuiiixQKhTRhwgSdfvrptVUjAADwqBoHlZtvvllfffWV+vTpo//5n//R+eefL7/fX+lUUDIxjKkAAOCoGgeVOXPmaMiQIbrmmmuS6gaEVbG4PhkAAFeo8SfTvvLKKzp48KCuvPJK/dd//ZcmT56sPXv21GZtjol+4Bs5BQAAZ9U4qGRnZ2vMmDH6/PPPddVVV2nmzJnKzc1VOBzWnDlzdODAgdqs80fFgAoAAO4Q971+6tatq8GDB2vKlCmaMWOGhg4dqueff149e/bUsGHDaqPGHx0f+AYAgDvEHVTKa9++ve655x598sknGj9+fKJqcg+GVAAAcNQPuilhlN/vV79+/dSvX79E7M5x0Q98I6cAAOCsHzSikqz4wDcAANyBoFKF2BQVkgoAAI4iqFSFybQAALgCQeUoGFABAMBZBJUqMJkWAAB3IKhUgc9RAQDAHQgqVSCnAADgDgSVYzCGE0AAADiFoFIFq9y5H3IKAADOIahUofypH3IKAADOIahUofxkWk79AADgHIJKFSym0wIA4AoElWNgPAUAAOck5O7Jx+vVV1/VlClTtG3bNklSx44ddfvtt6t3795OllVhkgpnfgAAcI6jQaVVq1YaMWKETj31VEnS9OnTdccdd+itt95Sx44dHaurwhwVx6oAAACOBpW+fftWeHzXXXdpypQpWrx4sbNBxbEjAwCA8hwNKuXZtq3/+7//U2FhobKzs+N+bmJrCR/+OhSSHfDGVJ7o65jo19PtvNi3F3uWvNm3F3uW6NvtfcdTn2Ucvv521apVuvrqq1VcXKy0tDQ99dRTNZ6jYtu2Fi9enPCaikrDum76TknSq1e0VKqfMRYAABItKytLfr//qNs4PqJy2mmnafr06dq/f79mz56te++9V5MnT9bpp59e431kZmYes9F4HCwOSdM/kCR16dxF9eumJGzfbmbbtpYuXZrw19PtvNi3F3uWvNm3F3uW6NvtfUfrrAnHg0pKSoratm0rKRI4li5dqn/+858aPXp0jffh9/sT+oYEAocHmXx+n6vf7NqQ6NfzROHFvr3Ys+TNvr3Ys0TfycB1ky+MMSopKXG0Bj7wDQAAd3B0RGX8+PHKzc1Vq1atdPDgQc2aNUvz58/XCy+84GRZFfA5KgAAOMfRoPL999/rnnvu0c6dO9WgQQNlZGTohRdeUK9evZwsi89RAQDAJRwNKo888oiTh68RRlQAAHCO6+aouIFVYYoKSQUAAKcQVKrAZFoAANyBoHIMnPoBAMA5BJUqMJkWAAB3IKhUofyJny/W7XasDgAAvI6gUgWr3JBKXv4hBysBAMDbCCrHUBrm5A8AAE4hqFSh/KmfrXsKHasDAACvI6hUofxk2tU7DzhXCAAAHkdQqUL5OSqldtjBSgAA8DaCyjGUhpijAgCAUwgqx9CyYarTJQAA4FkElWPw+fg4fQAAnEJQOYbiUtvpEgAA8CyCyjEcCjGZFgAApxBUjmFw99ZOlwAAgGcRVKrR9eSGkqRm9ZhMCwCAUwgq1Qj6I5NoQ3yOCgAAjiGoVCPgj7w0Nvf6AQDAMQSVavjLLkvmpoQAADiHoFKNYFlQYUQFAADnEFSq4fdFXhrmqAAA4ByCSjUCZSMqIUZUAABwDEGlGgE/p34AAHAaQaUascm0nPoBAMAxBJVqBLk8GQAAxxFUquFnjgoAAI4jqFQjNpnWJqgAAOAUgko1Dl/1wxwVAACcQlCpRvQj9Dn1AwCAcwgq1eDUDwAAziOoVMPPR+gDAOA4gko1oh/4xqkfAACcQ1CpxuFTP0ymBQDAKQSVagR8TKYFAMBpBJVqcOoHAADnEVSqEWAyLQAAjiOoVIObEgIA4DyCSjW4KSEAAM4jqFTDzwe+AQDgOIJKNaJzVHYWFDtcCQAA3kVQqcZna7+XJC3Zlu9wJQAAeBdBpRrvLtnhdAkAAHgeQaUGmFALAIAzCCrVGH1p59jXt7/ylYOVAADgXQSVavTv3DL29XvLv3OwEgAAvIugUo06Qb/TJQAA4HkElWo0qBNwugQAADzP0Z/GkyZN0uzZs7V+/XrVqVNH2dnZGjFihNq3b+9kWZWceUojp0sAAMCTHB1RmT9/vn71q19p6tSpevHFF2Xbtm688UYVFhY6WVYl32zls1QAAHCCoyMqf//73ys8HjdunM477zwtX75c55xzjkNVAQAAt3DVRIyCggJJUqNG8Z1qsW074bUcuc/aOIbbRHv0Qq/lebFvL/YsebNvL/Ys0bfb+46nPssY44pPMzPG6LbbbtP+/fv16quv1ug5tm1r8eLFtVbTVdN2KFT26rzxX61q7TgAAHhRVlaW/P6jX2XrmhGV0aNHa/Xq1TUOKeVlZmYes9F42batmzcUauLC/ZIiL2ays21bS5curZXX08282LcXe5a82bcXe5bo2+19R+usCVcElYcfflgfffSRJk+erFat4h+58Pv9tfKGnFQ/8vKkBnyufsMTrbZeT7fzYt9e7FnyZt9e7Fmi72Tg6FU/xhiNHj1as2fP1ssvv6w2bdo4WU4lobJ7/BSHwioJhR2uBgAA73E0qDz00EOaMWOGnnrqKdWrV0+7du3Srl27dOjQISfLijmlweEBp2c/WuNgJQAAeJOjQWXKlCkqKCjQkCFDlJOTE/sza9YsJ8uKaZZ2eNjsmY/WOlgJAADe5OgclVWrVjl5eAAA4HLc6+cYmtZLiX3NPBUAAH5cBJVjyCp3n5/73lziYCUAAHgPQeUYHr6sS+zrN7/e5mAlAAB4D0HlGFo1qlPhce7j/3GoEgAAvIegUgMjf54R+3rznkLtPlDsYDUAAHgHQaUG7uhzeoXH3cd84FAlAAB4C0Glhj68u7fTJQAA4DkElRrq0KJ+hceTv9zkUCUAAHgHQSUOGx+9JPb1A9OX6VCp7WA1AAAkP4LKD9Dpf/5Pew+WOF0GAABJi6ASp/KjKpKU/fD7DlUCAEDyI6gchyPDyobvDzpUCQAAyY2gcpzevys39nWfJz+WHTYOVgMAQHIiqBynji0bVHjcYdQs/YsrgQAASCiCyg+wZuxFFR7/z/RlanffTE1ftE3FIa4IAgDghyKo/ABBv0+/7Xt6peV3vr5YGQ/8n9rdN9OBqgAASB4ElR9oeP8MbXz0Er16c48q17e7b6b+d2nej1wVAADJgaCSID07NNe6Ry7W89efXWndba98rT99sNqBqgAAOLERVBLI77N0YeeW2vjoJZXmr/zpgzVqd99Mfb15r0PVAQBw4iGo1JKg36eNj16iP12VVWH5FX+Zq6smfaH8olJnCgMA4ARCUKllg7Jba9qw8yosm7dhj858aLba3TdTeflFDlUGAID7EVR+BGe3a6qNj16ihQ/0q7TuvHEfqd19M/l0WwAAqhBwugAvaV4/VeseuVjFIVudH3yvwro+T34sSTq/Y3P947/PUdBPhgQAgKDyI/P7LKWlBLTx0Us07autGvHvbyqs/2zN9+r4+/+VJHVq1UDf7ihQq4Z19O5vc9S8fqoTJQMA4BiCioMGdz9Fg7ufIjts1OORD/T9gZIK67/dUSBJ2rH/kM4e80GFdTflnKYberZTm6ZpP1q9AAD82AgqLuD3WVr4wIWSpKVb8zXw2c+P+ZwXPt+gFz7fEHt8WvN6OrVpmj5ZvUvntGuiW3I7KOf05qqb4q+1ugEAqG0EFZfJPKWRNj56SYVlhSUhvTR3oyxZ+tcXG7U9/1Cl5234/mBsQu6CjXu1YOPCCuvPa99M913USd1OaSTLsmqvAQAAEoigcgJISwno9gsi9xS67YIOseUr8/Zr0ifrNH3x9mPu44v1u3XZc3NUN+hXUamt05rX06Cs1up1ejNln9pEfh/hBQDgPgSVE9gZJzXUn67O1p+uzq60zhijzXsK9e6SPKX4ffrXl5u0eU+hikojd3Xe8P1BPf3Baj1dbupLk7Sg9haW6qdfzVe7ZvXUokGqmtZLUYsGqWpRP1UN6gTVOC2oFg1SuSoJAPCjIKgkKcuy1LZZPd3RJzISc3Nuex0oDmnDroP66Nuderrs3kOpAZ+KQ2FJ0t7CyKflfrl+j75cv+eo+29eP0WFJbbSUvzq1KqhmtZL0U8apKpJvRTVTw2oef1U1Uv1q3FaJOg0q5eiOkHmywAA4kNQ8ZD6qQFlntJImac00u/6dZQkldphrd91UDv2H9LWPQf17sJ16tKulYpDRt8fKFZJKKy8/EPKLypVflGpDpaEZIxiVygVltj6fO33NTp+g9SAAn5LLRvWUdDvU/P6KWrdpK4a1gmqfp2AGtQJqkX9FPl9PqUGfEpL8esnDeqocb2g6gb9jOIAgAcRVDwu6Pcpo1UDZbRqINtuqk7B3crK6iS/v+rRj5Ad1v5DIe0sOKSiElub9xRqV0GxCktsFRwq1e6DJTpYHNLuAyUqOBTS/kOl2llQLDtsVFAcknR45CZedYK+sr/9atWwjhrWDapBakCN0oJqkpaihnWCSg36VFhiyxgjY6QGZQEo8ndAPstSwGepYd2gUgI+1Qn4FbJtFRSHVVgSUv06PiYbA4CLEFQQl4Dfp6b1UtS0XookKfvUJsd8jjFG+4tC2nWgWIdKbe05WKJSO6xdBcXatq8oFmj2F5UqL/+QwibynIMlIe0qKNah0sipqfJ/7zvOsHNUMz5QnaBP9VODCoXDCvp9Cvos1a8TUOO6kX6NjCTJkqU6KX4FfZaCfp/qpviV4vcpJeCLhaOAz1KJHVaK3ycjo9SAX/VTA2pYNyhLkcvSUwM+Bfw+hcJhhSPtKei3VDfFr4Z1gmpYNzqaZCnAiBIADyKooNZZlqVGaUE1Sgse1/MPldo6VGorv6hUPstSYYmtbfsKVVhia39RSPuKSpRfFAk6h0ojwcDvtxT0WWUhKBKEDhwKyShyufem3YVK8ftUYoePOFZYh0qLKxaQf5yNJ5jfFxkN8lmWLCsyv8iyLJWGwjKKhDufZSkY8Mnvs2JhyJLK1kuWJdlho1I7rOLSkOrM/FB1UwJKCfgiASnol7/sGCkBn+oE/bHjSVLdoF9pKX7VCfplWZLPsuQr+9so8ndKWagKlgW3lIBPPkvyW5Z8Pkspfp8C/sgOA75ytZYt81mR906WVGob2WUJLjUQOaYlSwF/JCAG/ZYCvsj+fJaloN+SZR1+neywUZ2gTz6fJb9lyTJh2WGj4lBYvrAU8BEAAbcjqMD16gQjPxgbp6XElmW0apCw/du2rUWLFim9c6a+PxgJO0F/ZDQkZBsVHAopvygygmNZUihs5LMioSZkh1Vqh1VUaqvUNjpUaqvgUEgFh0IKG6OAz1IobGRJKg6Ftf9QqQoOhWSMkW2MikvDCoWN/NEfpJZUEgrrUKmt/YdCOlB2ukyKBAw7bGKPC0vsqhsqrnpxVQpLS6XaGJ1yuzdmx74M+KyyoOaLnTKMhK5I0Itcuh8JQX6fJWMi76UxJhaQonyWVRb0ygJXuRBkWYoFNJ8V2Y/PZylkhxV9W+sEffKXPS9QdjxLkbBfaocVLqtPOhxcA35f7HjRUoyRwsbIDkf62J+/T41XLJK/rEc7bGQpEmqDfkvBsrAXDZfRjqIBuPxrYoxkG6Nw2Ki07N92NLRGa/L7fGWh8nCYtcq9NpYVfT19ZaGy7PWLBnFFvs+MiYxhWtbh4G2HjcLGVAixkdfDp7KXN/KeGaOt2wq1OrRVfp8v9h74LEs+X/Q1Ohzwo99ZliSfLxKIdURv0b+jdUXDv9+yYu9t9P2Nso2J7TcaoqPP8ZX9ewqFw7F9WbJir53fFwne0X+L5vC3f4VjHe7HyITD2rCvVKl5++X3+2PPMzKxfR8pWlu01+j7JkkpAZ9aNapT+Uk/IoIKoMg3ab3UgBqmuet+SnY4En5KQmGVlIWicDjyn070VFiw3A/Lw/+Bm9gPtbAxsf+AwiYaiqQ1q75Vh/QMFYcizym1IwEpFI78ECqxwyoqsWP/IRsZFZXYKiqxVRyK/NAMl+1fUtkPwMgE7VA4HNtncWl020hdobBRSdmVZqV2ZDQoHDax/9DDZfuQIqcaU/yWwkYqDtmxfsImEugix4ocxxjFRsiioS76g+1oQmGjUIktqZrglyy2fed0Bc74apnTFTjj/bkJ29WoizvpltwOx96wlhBUABfz+yIBql6C85Nt2zq4PaCMlg2qnTh9ojNlwScUPhzcSkptfbNkic7slqlAIKDSUFjFoXAsCEZ/05UOjxrYZYEvZBuVhsOx3579Pksh21T4Ldc2kd/yjYmeYjOx347tsqAWHaWwFAlRfr8VO2ZRqR0bqbDDkRE9I0lGsWPGjhU2sZqkyOhAOHz4eNERjFDI1qYtW3TKKafIsiKjL9Ffq6MjgqW2qRD8oqMolnV4NMMqC7qx0YCykR8jU24E53AYjf2GXxYsjSJBNhw2MjKx9yXSo6kQQk1ZoI6OABxeZ2KjXNHRmOgrEioXdo2J7Hfvvn1q0KChZB0eBSn/OkVHoeywifUZfX/C5YJzNPRH38Pyr3FkdDTyWkaOLZWPxv6y0Y7o8uiIVrRXSbFTlWFjYs+NHTO6bVmNxpgKE/6j66O9GGNUXFIqfyAQW3943RHfI+W+V6Kve/Tv6Lq6Qb9aN3b2nnIEFQBJKfqfedBvKfoRPql+S/VTfGpQJxgJaO4aQKsVtm1rceoeZWW1TdpQWhXbtrV48WJlZWXR9wmOWWQAAMC1CCoAAMC1CCoAAMC1CCoAAMC1CCoAAMC1CCoAAMC1CCoAAMC1CCoAAMC1CCoAAMC1HA0qCxYs0LBhw5STk6OMjAx98MEHTpYDAABcxtGgUlhYqIyMDD344INOlgEAAFzK0Xv99O7dW71793ayBAAA4GJJcVNC20787dmj+6yNfbuVF3uWvNm3F3uWvNm3F3uW6NvtfcdTn2XMkTd+dkZGRoaee+459evXr8bPid4lEgAAnHhqcpfnE3pEJZqxOnfunPDbWdu2rRUrVtTKvt3Kiz1L3uzbiz1L3uzbiz1L9O32vqN11mSs5IQeUSkpKdHSpUtrsSoAAFBbMjMzlZKSctRtTugRlUAgoMzMTPl8PlmW5XQ5AACgBowxCofDCgSOHUMcDSoHDx7U5s2bY4+3bt2qlStXqlGjRjr55JOP+Xyfz3fMJAYAAE5cjp76mTdvnq6//vpKyy+//HI9+uijDlQEAADcxDVzVAAAAI7EvX4AAIBrEVQAAIBrEVQAAIBrEVQAAIBrEVQAAIBrEVQAAIBrEVSq8Morr6hv377KzMzUFVdcoYULFzpdUrUWLFigYcOGKScnRxkZGfrggw8qrDfGaMKECcrJyVG3bt00ZMgQrVmzpsI2JSUlevjhh9WjRw9lZWVp2LBh2rFjR4Vt8vPzNXLkSHXv3l3du3fXyJEjtX///grbbN++XcOGDVNWVpZ69OihMWPGqKSkJOE9T5o0SVdeeaWys7N13nnn6fbbb9f69euTuu9XX31VAwcO1FlnnaWzzjpLV111lT755JOk7bcqkyZNUkZGhsaOHRtblox9T5gwQRkZGRX+9OrVK6l7jvruu+80YsQI9ejRQ2eeeaYuu+wyLVu2LGl779u3b6X3OiMjQw899FBS9nvcDCqYOXOm6dKli5k6dapZu3atGTNmjMnKyjLbtm1zurQqffzxx2b8+PHmvffeM+np6eb999+vsH7SpEkmOzvbvPfee2bVqlXmzjvvNL169TIFBQWxbR588EFz/vnnmzlz5pjly5ebIUOGmEsvvdSEQqHYNjfeeKMZMGCA+frrr83XX39tBgwYYG699dbY+lAoZAYMGGCGDBlili9fbubMmWNycnLM6NGjE97zr3/9a/PGG2+Y1atXm5UrV5pbbrnFXHDBBebgwYNJ2/eHH35oPv74Y7N+/Xqzfv16M378eNOlSxezevXqpOz3SN98843p06ePGThwoBkzZkxseTL2/cwzz5hLLrnE7Ny5M/Zn9+7dSd2zMcbs27fP9OnTx9x3333mm2++MVu2bDFz5841mzZtStred+/eXeF9njNnjklPTzdffvllUvZ7vAgqRxg8eLB58MEHKyz7xS9+YZ588kmHKqq5I4NKOBw2vXr1MpMmTYotKy4uNt27dzdTpkwxxhizf/9+06VLFzNz5szYNjt27DCdOnUyn376qTHGmLVr15r09HSzePHi2DaLFi0y6enpZt26dcaYSGDq1KmT2bFjR2ybd99913Tt2rXCN1Vt2L17t0lPTzfz5883xnin73POOcdMnTo16fs9cOCA6d+/v5kzZ4657rrrYkElWft+5plnzKWXXlrlumTt2RhjnnjiCXPNNddUuz6Ze48aM2aM6devnwmHw57ot6Y49VNOSUmJli9frpycnArLe/XqpUWLFjlU1fHbunWrdu3aVaGflJQUnXPOObF+li1bptLS0gpDyy1btlTHjh1j2yxatEgNGjTQmWeeGdsmKytLDRo0iG2zePFidezYUS1btoxtk5OTo5KSkgpDt7WhoKBAktSoUSNJyd+3bduaOXOmCgsLlZ2dnfT9jh49Wr1791bPnj0rLE/mvjdt2qScnBz17dtXd911l7Zs2ZL0PX/00Ufq2rWrfvvb3+q8887ToEGDNHXq1Nj6ZO5divz8mTFjhq688kpZlpX0/cbjhL57cqLt3btXtm2rWbNmFZY3b95cu3btcqiq4xetuap+tm/fLkn6/vvvFQwGYz/ky2/z/fffx7Y5ch/R/Zbfpnnz5hXWN2rUSMFgMLZNbTDGaNy4cerevbvS09MlJW/fq1at0tVXX63i4mKlpaXpueee0+mnn66vv/46VteRvZzI/UrSzJkztWLFCk2bNq3SumR9n7t166bHHntM7dq10+7duzVx4kRdffXVevfdd5O2Z0nasmWLpkyZoqFDh2rYsGFasmSJxowZo5SUFA0aNCipe5ekDz74QAUFBbr88sslJe+/7+NBUKmCZVkVHhtjKi07kVTVz7HUdJvy+67uNarN12706NFavXq1Xn311WMe90Tv+7TTTtP06dO1f/9+zZ49W/fee68mT55c7fFO9H7z8vI0duxY/eMf/1Bqamq12yVb3717967wOCsrSxdeeKGmT58e+6042XqOHrtr164aPny4JKlz585au3atpkyZokGDBlV77GToXZLeeOMN5ebmVhjVqOp4ydJvPDj1U06TJk3k9/srJcjdu3dXSpsnghYtWkjSUftp3ry5SktLlZ+ff9Rtdu/eXWn/e/bsiSX1qkad8vPzVVpaWmWaT4SHH35YH330kV5++WW1atUqtjxZ+05JSVHbtm2VmZmpu+++W506ddI///nPpO13+fLl2r17t6644gp17txZnTt31vz58/Wvf/1LnTt3jtWdbH0fKS0tTenp6dq4cWPSvtdS5Pu2Q4cOFZa1b98+NnqQzL1v27ZNc+fO1eDBg2PLkrnfeBFUyklJSVGXLl00Z86cCsvnzp2r7Oxsh6o6fqeccopatGhRoZ+SkhItWLAg1k/Xrl0VDAYrbLNz506tWbMmtk12drYKCgq0ZMmS2DbffPONCgoKYttkZWVpzZo12rlzZ2ybOXPmKCUlRV27dk1oX8YYjR49WrNnz9bLL7+sNm3aeKLvIxljVFJSkrT9/vSnP9U777yj6dOnx/507dpVAwcO1PTp09WmTZuk7PtIJSUlWrdunVq0aJG077UknXXWWdqwYUOFZRs3blTr1q0lJff39ZtvvqlmzZrpggsuiC1L5n7j9mPN2j1RRC9P/ve//23Wrl1rxo4da7KysszWrVudLq1KBw4cMCtWrDArVqww6enp5sUXXzQrVqyIXU49adIk0717dzN79myzatUqM3z48Covb8vNzTVz5841y5cvN9dff32Vl7cNHDjQLFq0yCxatKjay9tuuOEGs3z5cjN37lyTm5tbK5e3/eEPfzDdu3c38+bNq3BpX1FRUWybZOv7qaeeMgsWLDBbtmwx3377rRk/frzp1KmT+fzzz5Oy3+qUv+onWft+9NFHzbx588zmzZvN4sWLza233mqys7Nj/wclY8/GRC5B79y5s5k4caLZuHGjmTFjhjnzzDPN22+/HdsmGXu3bdtccMEF5oknnqi0Lhn7PR4ElSpMnjzZ9OnTx3Tp0sVcfvnlscte3ejLL7806enplf7ce++9xpjIJX3PPPOM6dWrl+natav51a9+ZVatWlVhH4cOHTKjR4825557runWrZu59dZbzfbt2ytss3fvXnP33Xeb7Oxsk52dbe6++26Tn59fYZtt27aZW265xXTr1s2ce+65ZvTo0aa4uDjhPVfVb3p6unnjjTdi2yRb3/fff3/s3+RPf/pTc8MNN8RCSjL2W50jg0oy9h39rIwuXbqYnJwc85vf/MasWbMmqXuO+uijj8yAAQNM165dzS9+8Qvz+uuvV1ifjL1/9tlnJj093axfv77SumTs93hYxtRg1g0AAIADmKMCAABci6ACAABci6ACAABci6ACAABci6ACAABci6ACAABci6ACAABci6ACAABci6ACwDF9+/bVSy+95MixJ0yYoMsuu8yRYwOoOT6ZFkhy9913n/bv36+//OUvTpdSyZ49e1S3bl3VrVu3Vo+TkZGh5557Tv369YstO3jwoEpKStSkSZNaPTaAHybgdAEAkk9paamCweAxt2vatOmPUE3V6tWrp3r16jl2fAA1w6kfwOPWrl2rm2++WdnZ2erZs6dGjhypPXv2xNZ/+umnuuaaa3T22WerR48euvXWW7V58+bY+q1btyojI0OzZs3SkCFDlJmZqRkzZui+++7T7bffrr///e/KyclRjx499NBDD6m0tDT23CNP/WRkZOjf//637rjjDp155pnq37+/Pvzwwwr1fvjhh+rfv7+6deumIUOG6K233lJGRob2799fZX99+/aVJN1xxx3KyMiIPT7y1E+03r/+9a/q2bOnzj77bD377LMKhUJ67LHHdO655yo3N1fTpk2rsP/vvvtOd955p8455xz16NFDt912m7Zu3RrnuwCgOgQVwMN27typ6667TmeccYamTZumF154Qbt379add94Z26aoqEhDhw7VtGnT9NJLL8myLN1xxx0Kh8MV9vXkk09qyJAhmjVrlnJyciRJ8+bN0+bNm/Xyyy/r0Ucf1VtvvaW33nrrqDU9++yzuuiiizRjxgzl5uZqxIgR2rdvn6RIKPrd736nn/3sZ5o+fbquvvpqPf3000fdXzRYjBs3Tp9//nmloFHel19+qZ07d2ry5Mm67777NGHCBN16661q1KiRpk6dqquvvlp//OMflZeXF3ttrr/+eqWlpWny5Ml69dVXlZaWpptuukklJSVHrQtAzRBUAA+bMmWKunTpouHDh6tDhw7q3LmzHnnkEc2bN08bNmyQJP385z9X//791a5dO51xxhl65JFHtHr1aq1du7bCvm644Qb1799fbdq0UcuWLSVJjRo10oMPPqgOHTqoT58+6t27t7744ouj1nT55ZdrwIABatu2rYYPH66ioiItWbJEkvTaa6/ptNNO07333qv27dvrkksu0eWXX37U/UVPLzVs2FAtWrQ46ummxo0b64EHHlD79u01ePBgnXbaaTp06JCGDRumdu3a6dZbb1UwGNTXX38tSZo5c6Ysy9LYsWOVkZGhDh06aNy4ccrLy9P8+fOPWheAmmGOCuBhy5cv17x585SdnV1p3ebNm3Xaaadp8+bN+vOf/6zFixdr7969is6/z8vLU3p6emz7rl27VtrH6aefLr/fH3vcokULrV69+qg1ZWRkxL5OS0tTvXr1YqeiNmzYUOk43bp1q0GnNXP66afL5zv8+1vz5s3VsWPH2GO/36/GjRtr9+7dkiKv3+bNm3XWWWdV2E9xcXGF02MAjh9BBfCwcDisPn36aMSIEZXWtWjRQpI0bNgwnXTSSRozZox+8pOfKBwOa8CAARXmmkiRUHGkQKDifzGWZelYFxoeOQnXsqzYaSZjjCzLqrA+kRcuVlVvVcui9YTDYXXp0kVPPvlkpX05OVEYSCYEFcDDunTpovfee0+tW7eu9ANZkvbu3at169Zp9OjROvvssyVJCxcu/LHLjGnfvr0++eSTCsuWLVt2zOcFg0HZtp3werp06aL//d//VbNmzVS/fv2E7x8Ac1QATygoKNDKlSsr/Nm+fbuuvfZa5efna/jw4VqyZIm2bNmizz//XPfff79s21ajRo3UuHFjvf7669q0aZO++OILPfroo471cdVVV2nDhg164okntGHDBs2aNSs2OffIkZbyWrdurS+++EK7du1Sfn5+wuoZOHCgmjRpottuu00LFy7Uli1bNH/+fI0ZM0Y7duxI2HEALyOoAB4wf/58DRo0qMKfZ555Ri1bttSUKVMUDod14403asCAARo7dqwaNGggn88nn8+np59+WsuXL9eAAQM0btw43XPPPY710aZNG/35z3/W+++/r0svvVRTpkzRsGHDJEkpKSnVPu/ee+/V3LlzdcEFFxxz8m086tatq8mTJ+vkk0/Wb37zG1188cUaNWqUiouLGWEBEoRPpgVwQps4caJee+21SqeEACQH5qgAOKG88soryszMVJMmTfTVV1/p73//u371q185XRaAWkJQAXBC2bRpkyZOnKj8/HydfPLJGjp0qG699VanywJQSzj1AwAAXIvJtAAAwLUIKgAAwLUIKgAAwLUIKgAAwLUIKgAAwLUIKgAAwLUIKgAAwLUIKgAAwLX+H//dyV+w587PAAAAAElFTkSuQmCC"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 38
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-28T11:34:41.209401Z",
     "start_time": "2024-05-28T11:34:18.459032Z"
    }
   },
   "cell_type": "code",
   "source": [
    "arms = np.arange(3)\n",
    "loss = []\n",
    "A_map00 = {}\n",
    "b_map00 = {}\n",
    "for arm in arms:\n",
    "    A_map00[arm] = np.identity(len(list(val_set2)))\n",
    "    b_map00[arm] = np.zeros(len(list(val_set2)))\n",
    "\n",
    "for epoch in range(20):\n",
    "    # train\n",
    "    indices = np.random.permutation(list(range(len(val_set2))))\n",
    "    for i in indices:\n",
    "        feature = np.array(val_set2.iloc[i], dtype=float)\n",
    "        # iterate through actions\n",
    "        p = {}\n",
    "        max_value = -np.inf\n",
    "        max_arm = 0\n",
    "        for arm in arms:\n",
    "            theta = np.matmul(np.linalg.inv(A_map00[arm]), b_map00[arm])\n",
    "            p = np.matmul(theta.T, feature) + \\\n",
    "                np.sqrt(np.matmul(np.matmul(feature.T, np.linalg.inv(A_map00[arm])), feature))\n",
    "\n",
    "            if p == max_value:\n",
    "                if np.random.choice(a=[False, True]): max_arm = arm\n",
    "            elif p > max_value:\n",
    "                max_value = p\n",
    "                max_arm = arm\n",
    "\n",
    "        # observe reward and update A and b\n",
    "        r = -abs(val_dose2[i][0]-max_arm)\n",
    "        A_map00[max_arm] += np.outer(feature, feature)\n",
    "        b_map00[max_arm] += r * feature\n",
    "\n",
    "        if len(loss) == 0:\n",
    "            loss.append(-r)\n",
    "        else:\n",
    "            loss.append((loss[-1]*len(loss)-r)/(len(loss)+1))\n",
    "\n",
    "plt.plot(loss)\n",
    "plt.xlabel('Learning time')\n",
    "plt.ylabel('Average Loss')\n",
    "plt.show()"
   ],
   "id": "3ef7afc1f02aa712",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 37
  },
  {
   "metadata": {},
   "cell_type": "code",
   "outputs": [],
   "execution_count": null,
   "source": "",
   "id": "a0196701b76440a2"
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}