ML models predicting fare amounts, payment types, and demand patterns from NYC Yellow Taxi trip records

A machine learning project on NYC Yellow Taxi trip records. I built models to predict fare amounts, classify payment types, and forecast demand — useful for resource allocation and route planning.

Raw taxi data is messy. Getting useful models meant engineering features from temporal and geographic attributes, handling outliers in fare and trip duration, and pushing accuracy to where the predictions would actually be worth acting on.

Ran regression models (Linear, Lasso, Ridge, Elastic Net) reaching R² 0.825 on fare prediction, and classification models (Logistic Regression, Decision Trees, Random Forest, XGBoost) at 83.3% accuracy on payment type. Added time-series forecasting for demand from historical trip patterns.