🎯 Our goal: Accurately classify the fraudulent transactions from Xente's e-commerce platform and save money.

The data used for this is provided from this website: Xente Fraud Detection dataset.

Xente is an e-commerce and financial service app serving 10,000+ customers in Uganda. This dataset includes a sample of approximately 140,000 transactions that occurred between 15 November 2018 and 15 March 2019.

Our Evaluation Metric: F1-score

This challenge is a great exercise for classification problems and this notebook used various classification models, SMOTE sampling technique and Ensemble methods to improve the F1 score results.

📄 Contents:

• EDA and feature engineering
• Baseline model
• Logistic regression
• Decision tree
• Random forest
• SMOTE
• SMOTE with random forest
• Ensemble methods
• Loss prevented by using our final model
• Error Analysis

Requirements:

• pyenv with Python: 3.9.8

Environment:

For installing the virtual environment you can either use the Makefile and run make setup or install it manually with the following commands:

pyenv local 3.9.8
python -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install -r requirements.txt
pip install imbalanced-learn