🚗 Car Insurance Fraud Detection

This project aims to detect fraudulent car insurance claims using machine learning classification techniques. The dataset consists of various categorical and numerical variables representing customer, vehicle, and claim characteristics.

🧠 Objectives

• Identify potential fraudulent claims in a binary classification setup
• Handle class imbalance using SMOTE and alternative resampling techniques
• Compare model performance

🛠️ Tools & Techniques

• R
• Handling Imbalance: SMOTE, ROSE, Undersampling, Oversampling
• Models: Logistic Regression, SVM, Classification Tree, XG Boost, Random Forest
• Hyperparameter Tuning Methods: GridSearch with Cross Validation, Bayesian Optimization
• Evaluation metrics: Accuracy, Precision, Recall, F1 Score, ROC-AUC
• Model Interpretation: Partial Dependence Plots

📁 Project Structure

• /code/ – All R codes for preprocessing, modeling, and evaluation
• /report/ – PDF report and presentation slides
• /data/ – Includes the dataset

📄 License

Distributed under the MIT License. See LICENSE for more information.