Food Delivery Time Prediction

A machine learning solution to accurately predict food delivery times, improving customer experience and operational efficiency.

problem

Inaccurate ETA predictions were frustrating customers and leading to negative reviews. The existing rule-based system couldn't account for dynamic factors like traffic and weather.

approach

Trained a gradient boosting regression model incorporating real-time traffic data, weather conditions, restaurant preparation times, and historical delivery patterns. Implemented feature engineering for time-of-day and day-of-week effects.

outcome

Improved ETA prediction accuracy by 20% (MAPE reduced from 25% to 5%). Customer satisfaction scores increased by 15% as a result of more reliable delivery estimates.