Heart Disease Predictor

This project was my first real introduction to applied machine learning — not the theory of it, but the practical work of cleaning data, choosing the right model, and being honest about what 'accuracy' actually measures.

Three models were trained and compared: Random Forest, Logistic Regression, and Support Vector Machines. After tuning, Random Forest performed best, reaching approximately 88% accuracy. Users enter 13 clinical features and receive both a risk prediction and a confidence score, rather than a single unexplained number.

It taught me to be skeptical of my own models before I asked anyone else to trust them.