This project focuses on detecting fraudulent credit card transactions using various machine learning models. It leverages a real-world dataset with highly imbalanced classes, emphasizing precision, recall, and ROC-AUC for effective fraud identification.
The dataset contains transactions made by credit cards in September 2013 by European cardholders. It includes anonymized features (V1–V28), along with Time, Amount, and a Class label:
Class = 1: FraudClass = 0: Legitimate
Note: Due to confidentiality, the original dataset is not included. You can download it from Kaggle - Credit Card Fraud Detection
- Time: Seconds elapsed between each transaction and the first transaction in the dataset.
- V1–V28: Principal components obtained using PCA for confidentiality.
- Amount: Transaction amount.
- Class: Label indicating fraud.
-
Python (Pandas, NumPy, Scikit-learn, Matplotlib, Seaborn)
-
Jupyter Notebook
-
Machine Learning Models:
- Random Forest
-
Addressed class imbalance using evaluation metrics like Precision, Recall, F1-Score, and ROC-AUC.
-
Compared multiple models to identify the best performer for fraud detection.
-
Visualized confusion matrices and ROC curves for performance insight.
-
The importance of handling imbalanced datasets in classification problems.
-
How to evaluate models beyond accuracy, especially in domains where false negatives are critical.
-
Implementation and comparison of several classification algorithms.
-
How to use standardization techniques like
StandardScalerto improve model performance. -
Visualizing and interpreting results using confusion matrices and ROC curves.
-
The practical trade-offs between precision and recall when detecting rare events like fraud.