Minimizing the Sum of Squared Errors in Seasonally-Adjusted, Trend-Enhanced, Exponential-Smoothing Forecasting

Every business seeks to correctly anticipate how much of which products it must manufacture, and where to deliver them, to satisfy the requirements of its customers, as well as its own requirements for growth and profitability. Since the forecast of demand is the cornerstone of all other, subsequent planning, errors become very costly very quickly. This paper presents an optimization model for determining the smoothing constants and initial estimates of level, trend, and seasonality indices in Winter’s exponential smoothing forecasting model. The objective is to minimize the sum of squared forecast errors. An Excel template is available for download from the author’s website. The template has builtin macros to perform all calculations. Instructions on how to use the template are included in the template