需要予測における季節変動の統計的考察(第2報)

Abstract

The time series data analysis is one of the techniques used for demand forecasting. This analytical method consists of four types of movements such as trend movement, cyclical movement, seasonal movement and irregular component of demand. There have been several analytical methods available for dealing with seasonal movement. The previous report discussed the statisitical considerations on the seasonal distinction average method, link relatives method, moving average ratio method and sequential forecast method. The present study analyzes seasonal movement in a dummy valiable method and compares statistical considerations on the results of the previous report with those of the present study. The amount sold by the department stores were used to show numerical examples. As the results of caliculations, the probabilities for the forecast error at the levels of the ±5%, ±10% and ±15% forecasting accuracy are 0.65,0.91 and 0.98,respectively. The probabilities for the forecast error are the same or larger, compared with those of the previous report.