Built an end-to-end machine learning model to detect fraudulent credit card transactions using a real-world dataset containing over 284,000 records. The project involved cleaning the dataset, balancing the data using undersampling techniques to address class imbalance, and applying Logistic Regression to classify transactions as legitimate or fraudulent. Achieved an accuracy of ~92% with strong performance metrics (precision, recall, F1-score). The entire pipeline was implemented using Google Colab and Python libraries like pandas, NumPy, and scikit-learn. Tech Framework: Python, pandas, NumPy, scikit-learn, Google Colab, GitHub This project helped me understand the real-world challenges in fraud detection like class imbalance and model evaluation, while also improving my data preprocessing and model-building skills.