Uplift Modeling

• Author: Lina Faik
• Creation date: January 2024
• Last update: January 2024

Objective

This repository contains the code and the notebooks used to train and evaluate uplift models using CausalML and scikit-uplift libraries. It was developed as an experimentation project to support the explanation blog posts around the topic.

The model and code are explained in more detail in the following article:

• Beyond Predictions: Uplift Modeling & the Science of Influence
  Hands-On Approach to Uplift with Tree-Based Models (link coming soon)

You can find all my technical blog posts here.

Project Description

Code structure

data # folder containing the initial datasets
├── criteo-uplift-sample-v2.1.csv # sample of the data
notebooks 
├── 01_data_exploration.ipynb # data cleaning and exploration
├── 02_feature_engineering.ipynb # data preprocessing
├── 03_model_training.ipynb # uplift model training
├── 04_model_evaluation.ipynb # model evaluate and exploration
outputs # folder storing models and hyper-parameter research
src
├── training.py # general functions to train models           
├── viz.py # functions for data visualization

Data

The notebooks is based on criteo dataset that can be found in Kaggle. It can easily be adapted to other data adapted to uplift modeling

How to Use This Repository?

Requirement

The code uses a GPU and relies on the following libraries:

pandas==2.1.4
plotly==5.18.0
numpy==1.26.2
scikit-learn==1.3.2
scikit-uplift==0.5.1
optuna==3.5.0
causalml

Experiments

To run experiments, you need to run the notebooks in the order suggested by their names. The associated code is in the src directory.