Credit-Card-Fraud-Detection

This project focuses on detecting fraudulent credit card transactions using machine learning models. Since fraud cases are significantly fewer than legitimate transactions, an under-sampling technique is used to balance the dataset and improve model performance. Understood! Since your notebook uses undersampling instead of SMOTE, here’s the correct README file for your GitHub repository:

---

Dataset

• Source: (e.g., Kaggle’s "Credit Card Fraud Detection" dataset)
• Features: 28 principal components (PCA-transformed) + Time and Amount
• Target: Class (0 = Legitimate, 1 = Fraudulent)
• Imbalance Issue: Fraud cases account for <1% of total transactions

Techniques Used for Data Balancing

• Undersampling: Reducing the number of majority class samples (legitimate transactions) to match the minority class (fraudulent transactions).
• Why Undersampling? Helps prevent the model from being biased toward non-fraudulent transactions while maintaining a reasonable dataset size for training.

Technologies Used

• Python Libraries: Pandas, NumPy, Matplotlib, Seaborn
• Machine Learning: Scikit-learn (Logistic Regression, Decision Trees, Random Forest, etc.)
• Google Colab Features: GPU acceleration, Google Drive integration

Installation & Usage

1. Open the Colab notebook (Credit_Card_Fraud_Detection.ipynb).
2. Upload the dataset (if required).
3. Run the cells sequentially to preprocess data, apply undersampling, train models, and evaluate performance.

To use locally:

pip install -r requirements.txt

Then, open the notebook with:

jupyter notebook Credit_Card_Fraud_Detection.ipynb

Results

• Evaluation Metrics: Accuracy

---