README.md

Predicting Credit Card Fraud with Logistic Regression

Overview

This project, part of Codecademy's Machine Learning/AI Engineer path, predicts fraudulent credit card transactions using Logistic Regression.

Dataset & Features

The dataset (transactions_modified.csv) includes transaction details like amount, type, and account balances. Key engineered features:

• isPayment: 1 for DEBIT/PAYMENT, else 0
• isMovement: 1 for CASH_OUT/TRANSFER, else 0
• accountDiff: Difference between destination and origin balances

Model Training

• Data Split: 70% training, 30% test
• Scaling: StandardScaler normalizes features
• Classifier: Logistic Regression
• Evaluation: Model scores printed for training and test sets

Fraud Prediction

After training, the model predicts fraud in new transactions:

lr.predict(sample_transactions)
lr.predict_proba(sample_transactions)

Running the Code

Run the script in Python:

python fraud_detection.py

Future Improvements

• Try other ML models (e.g., Random Forest, Neural Networks)
• Tune hyperparameters
• Address class imbalance

Author

Codecademy Machine Learning/AI Engineer Path