forked from jesford/bike-sharing
-
Notifications
You must be signed in to change notification settings - Fork 0
/
bikerides.py
343 lines (250 loc) · 12.4 KB
/
bikerides.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
import numpy as np
import csv
import matplotlib.pyplot as plt
plt.ion()
print '\nEnter 1 to train and test on a subsample of the training data.'
print 'Enter 0 to train on full training sample, and test on the Kaggle test data.\n'
test_on_train = raw_input('ENTER 1 or 0: ')
#read in training data
csv_file_object = csv.reader(open('train.csv', 'r'))
header1 = csv_file_object.next()
data1 = []
for row in csv_file_object:
data1.append(row)
data1 = np.array(data1)
#optionally make the training set small even for applying to REAL
# test data for submission (random forests keep crashing...)
#temp = data1
#train_size = int(temp.shape[0]/5.)
#train_sample_indices = np.random.random_integers(0,(temp.shape[0]-1),train_size)
#data1 = temp[train_sample_indices,:]
if test_on_train == '1':
print '\nPreparing to train and test on subsets of Training Data...\n'
temp = data1
train_size = int(temp.shape[0]/20.)
train_sample_indices = np.random.random_integers(0,(temp.shape[0]-1),train_size)
data1 = temp[train_sample_indices,:]
test_sample_indices = [i for i in range(temp.shape[0]) if i not in train_sample_indices]
data2 = temp[test_sample_indices,:]
header2 = header1
true_count = temp[test_sample_indices, header1.index('count')].astype(int)
else:
print '\nPreparing to apply model to Kaggle Test Data...\n'
#read in testing data
test_file_object = csv.reader(open('test.csv', 'r'))
header2 = test_file_object.next()
data2 = []
for row in test_file_object:
data2.append(row)
data2 = np.array(data2)
#========================================
### MAIN FUNCTION BLOCK ###
def bikerides():
#select features to train on & organize data
train_data, train_count, test_data, test_datetime, important = feature_selection()
#select & run machine learning algorithm
learning_selection(train_data, train_count, test_data, test_datetime)
#plot features
print '------------------------------------\n'
wanttoplot = raw_input('Would you like to plot features? [enter y/n]: ')
while wanttoplot == 'y':
plotting_selection(important, train_data, train_count)
print '\n------------------------------------'
wanttoplot = raw_input('\nWould you like to make another plot? [enter y/n]: ')
#plt.close()
plt.figure() #start new figure, leave others up
print '\nFinished\n'
return
### end bikerides MAIN block
#========================================
# select which features to use and set up arrays...
def feature_selection():
print '------------------------------------\n'
choose_features = raw_input('CHOOSE IMPORTANT FEATURES TO TRAIN ON.\nYou can select any of the following features:\n\n time \n season \n holiday \n workingday \n weather \n temp \n atemp \n humidity \n windspeed \n\nList names separated by a space, for example: weather atemp \n (you can also enter "ALL") \n\nENTER FEATURES: ')
if choose_features == 'ALL':
important = ['time', 'season', 'holiday', 'workingday', 'weather', 'temp', 'atemp', 'humidity', 'windspeed']
else:
important = choose_features.split()
print '\nOrganizing data structures.\n'
train_data = np.zeros([data1.shape[0],len(important)])
test_data = np.zeros([data2.shape[0],len(important)])
for j, item in enumerate(important):
print '...processing: ', j, item
if item == 'time':
time1 = [datetime[11:13] for datetime in data1[:,0]] #pick out hours
train_data[:,j] = np.array(time1).astype(np.float) #convert to float
time2 = [datetime[11:13] for datetime in data2[:,0]]
test_data[:,j] = np.array(time2).astype(np.float)
else:
#get corresponding indices for each data set
itrain = header1.index(item)
itest = header2.index(item)
#convert all entries to float
train_data[:,j] = data1[:,itrain].astype(np.float)
test_data[:,j] = data2[:,itest].astype(np.float)
train_count = data1[:,-1].astype(np.float)
test_datetime = data2[:,0]
return train_data, train_count, test_data, test_datetime, important
### end feature_selection
#========================================
# choose a machine learning method
def learning_selection(datatrain, alivetrain, datatest, datetime):
print '\n------------------------------------'
print '\nChoose a Machine Learning technique from the following list: \n'
print ' 1. ORDINARY LEAST SQUARES \n\t fits a linear model by minimizing the residual sum of squares'
print ' 2. STOCASTIC GRADIENT DESCENT (SGD) \n\t linear classifier applied to a normalized/standardized \n\t version of the data, using the hinge-loss option for penalties; \n\t searches can be more efficient using gradient information'
print ' 3. BAYESIAN RIDGE REGRESSION \n\t fits a linear model by maximizing the marginal log(Likelihood);\n\t can be more robust to poorly-defined problems'
print ' 4. RANDOM FOREST \n\t each tree gets a bootstrapped sample of training data, and \n\t branches are chosen that yield the best outcome for a random \n\t subsample of features; final result is the model average of \n\t all 100 trees'
print ' 5. EXTREMELY RANDOMIZED FOREST \n\t similar to above, but sets a random threshold for whether a \n\t branch outcome is considered better or not; can further \n\t reduce variance, but may be more biased'
print ' 6. SUPPORT VECTOR MACHINE (SVM) \n\t multi-class classification using subsets of training data \n\t (support vectors); can be effective in many dimensions, \n\t usable with more features than training samples'
print ' 7. NAIVE BAYES \n\t assumes features are independent and gaussian; fast to run \n\t and can be trained on very small samples'
print ' 8. BERNOULLI NAIVE BAYES \n\t assumes binary distributions of data, or may manipulate data \n\t into this form'
print ' 9. ADABOOST \n\t ensemble method (like forests) that uses weights on the \n\t training samples to boost importance of incorrect predictions, \n\t so that improvements can be made before outputting the average \n\t of all 100 weak learners'
print ' 10. GRADIENT BOOSTED REGRESSION TREES (GBRT) \n\t another ensemble method with 100 weak learners; robust to \n\t outliers and handling of mixed data types'
choose_method = int(raw_input('\nENTER THE # OF THE TECHNIQUE YOU WANT TO APPLY: '))
#-------------
# ORDINARY LEAST SQUARES
if choose_method == 1:
print '\nRunning OLS...\n'
from sklearn import linear_model
ols = linear_model.LinearRegression()
ols.fit(datatrain,alivetrain)
Output = ols.predict(datatest)
#-------------
# STOCASTIC GRADIENT DESCENT (SGD)
elif choose_method == 2:
print '\nRunning SGD Classifier...\n'
from sklearn.linear_model import SGDClassifier
#normalize feature scaling (SGD is sensitive to this)
# note: this helps significantly (~10% improvement in score)
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
scaler.fit(datatrain)
datatrain = scaler.transform(datatrain)
# apply same transformation to test data
datatest = scaler.transform(datatest)
sgdc = SGDClassifier(loss="hinge", penalty="l2")
sgdc.fit(datatrain,alivetrain)
Output = sgdc.predict(datatest)
#-------------
# BAYESIAN RIDGE REGRESSION
elif choose_method == 3:
print '\nRunning Bayesian Ridge Regression...\n'
from sklearn import linear_model
brr = linear_model.BayesianRidge()
brr.fit(datatrain,alivetrain)
Output = brr.predict(datatest)
#-------------
# RANDOM FOREST
elif choose_method == 4:
print '\nRunning Random Forest Classifier...\n'
from sklearn.ensemble import RandomForestClassifier
Forest = RandomForestClassifier(n_estimators = 100) #1000 trees
Forest = Forest.fit(datatrain,alivetrain)
Output = Forest.predict(datatest)
#-------------
# EXTREMELY RANDOMIZED FOREST
elif choose_method == 5:
print '\nRunning Extremely Randomized Forest...\n'
from sklearn.ensemble import ExtraTreesClassifier
extratrees = ExtraTreesClassifier(n_estimators = 100) #1000 trees
extratrees = extratrees.fit(datatrain,alivetrain)
Output = extratrees.predict(datatest)
#-------------
# SUPPORT VECTOR MACHINES
elif choose_method == 6:
print '\nRunning SVM Classifier...\n'
from sklearn import svm
clf = svm.SVC()
clf.fit(datatrain,alivetrain)
Output = clf.predict(datatest)
#-------------
# NAIVE BAYES
elif choose_method == 7:
print '\nRunning Gaussian Naive Bayes...\n'
from sklearn.naive_bayes import GaussianNB
gnb = GaussianNB()
gnb.fit(datatrain,alivetrain)
Output = gnb.predict(datatest)
#-------------
# BERNOULLI NAIVE BAYES
elif choose_method == 8:
print '\nRunning Bernoulli Naive Bayes...\n'
from sklearn.naive_bayes import BernoulliNB
bern = BernoulliNB()
bern.fit(datatrain,alivetrain)
Output = bern.predict(datatest)
#-------------
# ADABOOST
elif choose_method == 9:
print '\nRunning AdaBoost Classifier...\n'
from sklearn.ensemble import AdaBoostClassifier
ada = AdaBoostClassifier(n_estimators=100)
ada.fit(datatrain,alivetrain)
Output = ada.predict(datatest)
#-------------
# GRADIENT TREE BOOSTING
elif choose_method == 10:
print '\nRunning GBRT Classifier...\n'
from sklearn.ensemble import GradientBoostingClassifier
grad = GradientBoostingClassifier(n_estimators=100)
grad.fit(datatrain,alivetrain)
Output = grad.predict(datatest)
#----------------------------------------
# Either analyze response to the subset of Training Data OR
# output result on Kaggle Test Data to a file for submission...
count = Output.astype(np.int)
if test_on_train == '1':
print '------------------------------------\n'
print 'Comparing to known counts.\n'
#OUTPUT SOME STATISTICAL COMPARISONS HERE!
#print true_count.shape,count.shape, len(true_count), len(count)
#print 'LESS THAN ZERO?? ', len(np.where(count < 0.))
count[np.where(count < 0.)] = 0.
n = len(true_count)
summation_arg = (np.log(count+1.) - np.log(true_count+1.))**2.
rmsle = np.sqrt(np.sum(summation_arg)/n)
print 'RMSLE', rmsle, '\n'
else:
print 'Saving Predictions in file: output.csv\n'
f = open('output.csv', 'w')
open_file_object = csv.writer(f)
open_file_object.writerow(['datetime','count'])
for i in range(len(datetime)):
open_file_object.writerow([datetime[i],count[i]])
f.close()
return
### end of learning_selection()
#========================================
# Plotting option function
def plotting_selection(important, train_data, train_count):
print '\nCHOOSE FEATURES TO PLOT.\nYou can select any one of the features that you used for analysis. \nCOUNT will be plotted as a function of your variable.\n'
for j,item in enumerate(important):
print str(j+1)+'. '+item
plotfeatures = raw_input('\nENTER THE # of A FEATURE: ')
index = int(plotfeatures) - 1
x_var = train_data[:,index] #x-axis variable
y_var = train_count #y-axis variable
jitter = np.random.randn(len(y_var))/2.
if important[index] in ['season','holiday','workingday','weather']:
jitter = jitter/5. #smaller jitter
plt.scatter(x_var+jitter,y_var,color='darkcyan', marker='+')
if important[index] in ['holiday','workingday']:
ix = np.digitize(x_var,bins=[-0.5,0.5,1.5])
elif important[index] in ['season','weather']:
ix = np.digitize(x_var,bins=[0.5,1.5,2.5,3.5,4.5])
else:
b = [i for i in range(int(np.min(x_var)-1),int(np.max(x_var)+1),3)]
ix = np.digitize(x_var,bins=b)
for i in list(set(ix)):
here = np.where(ix == i)
x_mean = np.mean(x_var[here])
y_mean = np.mean(y_var[here])
plt.scatter(x_mean,y_mean,color='black', marker='o', s=100)
plt.xlabel(important[index])
plt.ylabel('count')
#plt.title('blue = lived, orange = died')
return
### end of plotting_selection
####################################
bikerides()