AI DELIVERY TIME PREDICTION SYSTEM

Abstract

The AI Delivery Time Prediction System is designed to estimate the delivery time of food orders using machine learning techniques, aiming to improve efficiency and customer satisfaction in logistics and food delivery services. The system utilizes a dataset of 45,593 food delivery records with 20 features, including delivery personnel details, vehicle type and condition, weather conditions, traffic density, distance between locations, festival occurrences, city type, and order type. Additionally, a new feature called Haversine distance is engineered using GPS coordinates to calculate the real-world distance between the restaurant and the delivery location, enhancing prediction accuracy. Multiple machine learning regression models, including Linear Regression, Decision Tree Regressor, Random Forest Regressor, and Gradient Boosting Regressor, are implemented and evaluated. The models are assessed using standard performance metrics such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), R² Score, and Mean Absolute Percentage Error (MAPE). Among these, the Gradient Boosting Regressor achieves the best performance with a high R² score of 0.89, along with low error values, indicating strong prediction accuracy. The reliability of the model is further validated using 5-fold cross-validation. The system is deployed as an interactive desktop application developed using Python’s Tkinter library, featuring a user-friendly graphical interface. Users can input order details and receive instant delivery time predictions without system delays. Additionally, the application includes a PDF report generation feature for documenting prediction results.