A Strategic Analysis of Food Demand Using Machine Learning and Explainable AI

The growing importance of meal demand forecasting is increasingly recognized as vital for business growth, aiding in efficient procurement planning and minimizing waste. This paper employs the 'Meal demand forecasting' dataset from Kaggle to analyze customer order patterns at the top ten profit-making fulfillment centers, aiming to predict future food item demand. This focus enhances the precision of our analysis and underlines the need for optimized inventory management in key areas. Various regression models, including Random Forest, Xgboost, Linear, Ridge, and Lasso Regressors etc. were tested. The models were assessed through the examination of metrics such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and R-squared to provide a comprehensive evaluation. Furthermore, this study also addresses model interpretability, a significant gap in current research, by explaining how different features impact demand predictions. This effort aims to provide more understandable and actionable insights for strategic decision-making in meal demand forecasting.