Product Specific Forecasting and Capacity Allocation: A Hybrid Model for Profit Optimization in Multi Variant Food Manufacturing

Abstract

This study addresses production planning challenges in food manufacturing SMEs characterized by heterogeneous demand patterns, limited historical data, and fixed capacity constraints. A hybrid model integrating product-specific demand forecasting and profit-maximizing linear programming is proposed to generate optimal annual production targets and profit projections. Historical sales data for four product variants are first preprocessed using the Interquartile Range (IQR) method to remove non-representative outliers. Demand patterns are then diagnosed as Beta-like, Exponential-like, or Triangular, guiding the selection of tailored forecasting methods: Moving Average (MA) for Product A, Single Exponential Smoothing (SES, α=0.5) for Products B and C, and Simple Average (SA) for Product D. The resulting forecasts serve as minimum production constraints in a linear programming model that maximizes total contribution profit under a 72,000-unit annual capacity. Implemented in POM-QM, the model yields a projected annual profit of IDR 213,514,600, with sensitivity analysis revealing a 7.7% improvement over volume-based planning and a shadow price of IDR 3,388/unit for capacity up to 75,000 units. This research demonstrates that operational excellence in resource-constrained environments stems not from model complexity, but from the thoughtful alignment of foundational quantitative methods with empirical demand behavior and financial logic.