Forecasting methods and seasonal adjustment for a university foodservice operation

Considering the unique seasonal pattern in university dining environments, this study attempts to determine the degree of improvement in accuracy of each forecasting method tested when seasonally adjusted data are employed. This study also seeks to identify the most accurate forecasting method of the six forecasting methods used in this study: naïve, moving average, simple exponential smoothing, Holt's method, Winter's method, and linear regression. Accuracy is measured using Mean Squared Error, Mean Absolute Percentage Error, and Mean Percentage Error. Results show that Winter's method outperforms the other five methods when raw data are used, while Moving Average method, when used with seasonally adjusted data, is the most accurate forecasting technique. Seasonally adjusted data are found to greatly improve forecasting accuracy in most of the methods. The findings of this study indicate that seasonally adjusted data are more effective in forecasting customer counts in the university foodservice operations than raw data, so the adjusted data help control costs and increase customer satisfaction.