Using the calendar to improve sales forecasts

Journal of the Academy of Marketing Science · 1984 · Volume 12 · Issue 3

Abstract

Monthly sales data for products whose sales fluctuate regularly according to the day of the week may be subject to a source of variation which makes model fitting and forecasting more difficult. This problem is described and illustrated within the context of the ARIMA class of univariate models. Procedures to recognize and account for the problem are presented, and improvements in the ability to forecast future sales are noted in the sample series examined.

Keywords

Forecast Error, Mean Absolute Percent Error, ARIMA Model, Retail Sale, Monthly Time Series

Enable JavaScript for citation tools and additional paper interactions.