Browse Prior Art Database

Computer Calculation of Promotional Intensities

IP.com Disclosure Number: IPCOM000077486D
Original Publication Date: 1972-Aug-01
Included in the Prior Art Database: 2005-Feb-25
Document File: 3 page(s) / 23K

Publishing Venue

IBM

Related People

Nigro, PD: AUTHOR

Abstract

This application program forecasts promotional sales based on the behavior of past promotions. Forecast accuracy is increased by specifying intensities for each promotion, and by using a computer method of determining intensities for each historical promotion. Historical sales information with the identification of when in the historical sales history the promotion(s) occurred is required as input. Standard techniques remove seasonality and trend from historical sales information. For identification and removal of trend, remove the seasonal influence by dividing the historical sales by a seasonal indices vector; and using deseasonalized sales, compute a data string of cumulative sums where each element of the data string consists of the sum of historical sales before and including that time period.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 57% of the total text.

Page 1 of 3

Computer Calculation of Promotional Intensities

This application program forecasts promotional sales based on the behavior of past promotions. Forecast accuracy is increased by specifying intensities for each promotion, and by using a computer method of determining intensities for each historical promotion. Historical sales information with the identification of when in the historical sales history the promotion(s) occurred is required as input. Standard techniques remove seasonality and trend from historical sales information. For identification and removal of trend, remove the seasonal influence by dividing the historical sales by a seasonal indices vector; and using deseasonalized sales, compute a data string of cumulative sums where each element of the data string consists of the sum of historical sales before and including that time period. That is, if deseasonalized historical sales is

sales [1], sales [2],..., sales [N] (1)

then the cumulative sum data string is

CS[1], CS[2],..., CS[N] where CS[t] = (see orig. page 772) sales
[j] (2). A quadratic equation or a step function is fitted to the CS data string. This fitted curve estimates trend.

By subtracting the fitted curve in #2 above from deseasonalized sales during the promotion time periods, sales due to a promotion is now isolated.

Promotions are made to correspond to each other by relating each promotion to four different periods: (1) The time the promotion is announced; (2) the time the promotion begins; (3) the time the promotion ends; (4) the time the postpromotional effect stops. The set of all promotions for a particular item is P = {Rho(ij)}, where j represents the promotion number and i the time period relative to the announce time.

Based on the foregoing, the user specifies the relative intensities of a set of promotions, for a particular item using the computer calculation of intensities.

Computer calculation of intensities can be explained using the following example, assuming there are three promotions for an item and the promotions are each eight periods long as follows: Promotion Period/Value 1 2 3 4 5 6 7 8

1 10 20 30 40 50 40 20 10

2 15 15 80 75 50 30 25 15

3 20 40 100...