AI Driven Time Series Forecasting For Food Supply Chain Optimization

Abstract

Perishable food supply chains are exposed to pronounced demand volatility driven by promotions, weather, and festival effects; consequently, traditional statistical forecasters often yield biased or lagged signals that propagate into stockouts, overstocking, and avoidable waste. This work presents an integrated decision-support framework that couples multi-horizon, exogenous-aware time series forecasting with inventory optimization tailored to perishable goods. The forecasting layer benchmarks classical models against machine learning (gradient-boosted trees, ensembles) and deep learning architectures (LSTM/GRU, Temporal Fusion Transformer, PatchTST), explicitly incorporating external covariates to capture non-linear and non-stationary demand regimes. Forecast distributions then feed an optimization layer that applies the Economic Order Quantity model for relatively stable items and the Newsvendor formulation, as well as linear/mixed-integer programs, for short-life products across SKU–store hierarchies. The system is evaluated using statistical accuracy metrics (MAPE, RMSE, Bias) and operational key performance indicators (service level, fill rate, holding cost, and wastage percentage). An interactive dashboard operationalizes these components, enabling scenario analysis and proactive alerts for stockout or overstock risk. By jointly improving forecast fidelity and translating predictions into implementable replenishment rules, the framework targets measurable reductions in waste and cost while sustaining customer service levels in real-world retail contexts.