Vehicle Crash Risk Prediction
Built a machine learning pipeline using NHTSA’s FARS 2024 national crash dataset to predict crash injury severity and generate continuous 0–100 insurance risk scores Trained a RandomForestClassifier with class-weight balancing to address class imbalance, using predict proba() outputs to drive a premium-multiplier scoring function Structured the project into modular components (preprocessing, training, prediction) and resolved dtype/environment issues in a Jupyter/VS Code workflow