DEMAND ESTIMATION DEVICE AND PROGRAM

摘要

PROBLEM TO BE SOLVED: To improve the estimation accuracy in multiple regression analysis in consideration of effect of the number of commodities displayed (display state) in a shop to a purchase estimation.SOLUTION: A demand estimation device has an estimated purchase quantity calculation part for calculating an estimated purchase quantity per commodity included in a commodity group within a second prescribed period which is a period after a first prescribed period by using a non-self sum regression coefficient indicating a degree of contribution of the sum of purchase quantities of non-self commodities other than an object commodity which is a prescribed commodity included in the commodity group to variation in purchase quantity of the object commodity, a variation regression coefficient indicating a degree of contribution of variation factor information of the object commodity to variation in purchase quantity of the object commodity and the variation factor information of the object commodity. The estimated purchase quantity calculation part calculates a regression curve statistically approximating a curve of a purchase quantity by using a time zone as the variation factor information, and, on the basis of an averaged regression curve for an average purchase quantity, determines that there is an abnormality in a commodity display state when the gradient of the regression curve of the purchase quantity is smaller than the averaged regression curve for each time zone so as to correct the estimated purchase quantity.