UI:Streamlit
Platform:Google Colab\
This project aims to detect fraudulent transactions in the SecureSwipe dataset. The dataset contains both fraudulent and successful transactions, and the goal is to build a model that can accurately classify transactions as fraudulent or successful.
- TensorFlow
- Keras
- NumPy
- Pandas
- Matplotlib
- Seaborn
- scikit-learn
- scipy
- imbalanced-learn
- tabulate
- plotly
data/
: Contains the dataset used for training and testing.src/
: Contains the source code for data preprocessing, model training, and evaluation.models/
: Contains saved trained models.results/
: Contains evaluation results and visualizations.
- Removed null, missing, and duplicate values.
- Scaled 'amt' and 'time' columns using RobustScaler to handle outliers.
- Detected and handled outliers using the Interquartile Range (IQR) method.
- Oversampled the data using Synthetic Minority Over-sampling Technique (SMOTE) due to class imbalance.
- Utilized Logistic Regression as the classification model.
- Checked for multicollinearity using Variance Inflation Factor (VIF).
- Addressed multicollinearity by applying Ridge regularization.
- Evaluated the model using the following metrics:
- Accuracy
- Precision
- Recall
- Confusion Matrix
- Implemented Privacy-Preserving Aggregated Training with Exponential Mechanism (PATE) using Laplacian noise for security.
- Clone the repository:
git clone https://github.com/your-username/SecureSwipe-Fraud-Detection.git
Navigate to the project directory:
cd SecureSwipe-Fraud-Detection
Install dependencies:
pip install -r requirements.txt