Method for classifying sale amount characteristics, method for predicting sale volume, method for ordering for restocking, system for classifying sale amount characteristics and system for ordering for restocking

• US 5,596,493 A
• Filed: 12/17/1992
• Issued: 01/21/1997
• Est. Priority Date: 04/19/1991
• Status: Expired due to Fees

• Current Assignee: Meiji Milk Products Company Limited and Kaoru Tone
• Original Assignee: Kaoru Tone, Meiji Milk Products Co Ltd (Meiji Holdings Company Limited)
• Inventors: Tanaka, Hiroyuki, Okazaki, Toshikazu, Tone, Kaoru, Tone, Michiko
• Primary Examiner(s): McElheny, Jr., Donald E.
• Assistant Examiner(s): VERDUN, HAYWARD AUBRY
• Application Number:
  US07/956,779
• Time in Patent Office: 1,496 Days
• Field of Search: 364/403, 364/404, 364/405, 364/401, 364/400, 235/375-385
• US Class Current: 705/7.31
• CPC Class Codes: G06Q 10/087 Inventory or stock management, e.g. order filling, procurement or balancing against orders
• CPC Class Codes: G06Q 20/203 Inventory monitoring
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities
• CPC Class Codes: G07G 1/145 PLU-management

Abstract

POS data are entered at step S1. The input POS data are arranged and those data for a day on which there was no stock at the store opening time and for a day on which the goods are out of stock at the store closure time are discarded to formulate data sets of the daily sale amounts for individual goods. At step S3, non-routine goods are discarded. At step S4, basic statistic values of the goods, such as mean value, standard deviation, maximum value, minimum value, skewness value, kurtosis value, Geary value etc. of the daily sale amounts of the goods are calculated. At step S5, the goods are classified into one of preset plural types, such as Poisson type, normal type, causal type and other type. Besides, an optimum amount for restocking order is found on the basis of the class types.

Claims

1. A sale amount prediction method comprising the steps of:

entering POS data into a point-of-sale terminal;

calculating, within a computer and based upon the entered POS data, sale amount data of the individual goods for a predetermined time period;

calculating, within the computer and based upon the calculated sale amount data, a plurality of basic statistic parameters of daily sale amounts of said individual goods;

classifying, within the computer and based on said calculated basic statistic values, said individual goods by determining into which of a Poisson type, a normal type, a causal type, and another type not belonging to one of these types, the daily sale amount of the individual goods belongs; and

predicting the sale amount of the individual goods based upon a characteristic value of the type into which the daily sale amount of the individual goods is classified.

2. The sale amount prediction method according to claim 1 wherein, prior to calculation of the basic statistic parameters, data of individual goods for a day when such individual goods are already depleted on store opening and data of individual goods for a day when such individual goods are depleted on store closure are discarded from the sale amount data of the individual goods and the basic statistic parameters of the daily sale amount for the individual goods are calculated based on the daily sale amount data of the individual goods after such discarding.

3. The sale amount prediction method according to claim 1 wherein, prior to calculation of the basic statistic parameters, data of goods not displayed routinely and data of scrapped goods are discarded and the basic statistic parameters of the daily sale amount for the individual goods are calculated based on the daily sale amount data of the individual goods after such discarding.

11. The sale amount prediction method according to claim 1 wherein the plurality of basic statistic parameters includes

a mean value, a standard deviation, a maximum value, a minimum value, a skewness value, a kurtosis value and a Geary value.

12. The sale amount prediction method according to claim 1 wherein the classifying step includes:

determining by linear multiple regression formula employing said basic statistic parameters, whether the daily sale amount of the individual goods fits a Poisson distribution and, if the result is affirmative, classifying the daily sale amount of the individual goods as the Poisson type;

otherwise, determining by chi-square testing employing the said basic statistic parameters, whether the daily sale amount of the individual goods fits a normal distribution and, if the result is affirmative, classifying the daily sale amount of the individual goods as a normal type;

otherwise, determining whether a coefficient of each casual in a linear multiple regression formula employing the daily sale amount of the individual goods is larger than a pre-set value and, if the result is affirmative, classifying the daily sale amount of the individual goods as a casual type; and

otherwise, classifying the daily sale amount of the individual goods as an other type.

13. The sale amount prediction method according to claim 1 wherein the predicting step includes

predicting the sale amount of the individual goods with said characteristic value being a mean value when the type into which the daily sale amount of the individual goods is the Poisson type,

predicting the sale amount of the individual goods with said characteristic value including a mean value and a standard deviation when the type into which the daily sale amount of the individual goods is the normal type,

predicting the sale amount of the individual goods with said characteristic value including a linear multiple regression formula when the type into which the daily sale amount of the individual goods is the causal type, and

predicting the sale amount of the individual goods with said characteristic value including a mean value and standard deviation when the type into which the daily sale amount of the individual goods is the other type.

14. The sale amount prediction method according to claim 1, wherein the normal type is one of

a normal type based only upon the sale amount,

a normal type based upon the sale amount for preset numbers of guests,

a lognormal type based upon the log of the sale amount and

a normal type based upon the log of the sale amount for preset numbers of guests.

4. A sale amount prediction system comprising:

POS data entry means for entering POS data;

sale amount data calculating means for calculating sale amount data of individual goods over a predetermined time period based on POS data not entered via said POS data entry means;

basic statistic parameter calculating means for calculating basic statistic parameters of daily sale amounts of said individual goods based on the sale amount data of said individual goods calculated by said sale amount calculating means; and

means for classifying said individual goods by verifying, based upon the basic statistic parameters calculated by said basic statistic parameter calculating means, into which of a Poisson type, a normal type, a casual type, and an other type not belonging to these types, belongs the daily sale amount of the individual articles, to classify the individual goods; and

sale amount prediction means for predicting a sale amount of the individual goods based upon characteristic values of the type into which the individual goods are classified by said classifying means.

5. The sale amount prediction system according to claim 4, further including

means for calculating the amount of a restocking order based on the estimated sale amount and an amount on stock.

6. The sale amount prediction system according to claim 5 wherein said means for calculating the amount of a restocking order is provided in a hand-held terminal for inputting a stock amount of each individual goods, data of the estimated sale amount from said means for estimating the sale amount being transmitted to said means for calculating the amount of the restocking order for calculating the amount of the restocking order, and wherein the results of calculation are displayed on a display part of said hand-held terminal.

7. The sale amount prediction system for classifying individual goods according to claim 4 wherein, prior to calculation of the basic statistic parameters, data of individual goods for a day when such individual goods are already depleted on store opening and data of individual goods for a day when such individual goods are depleted on store closure are discarded from the sale quantity data of the individual goods and the basic statistic parameters of the daily sale amount for the individual goods are calculated based on the daily sale amount data of the individual goods after such discarding.

8. The sale amount prediction system according to claim 4, wherein the sale amount prediction means classifies the daily sale amount of the individual goods into one of

a normal type which depends only upon the sale amount,

a normal type which depends on the sale amount per a predetermined unit number of guest,

a lognormal type which depends on the logarithm of the sale amount;

and

a lognormal type which depends on the logarithm of the sale amount per a predetermined unit number of guests.

9. The sale amount prediction system according to claim 4, wherein the classifying means includes:

Poisson type distribution means for determining, by a linear regression formula employing the basic statistic parameters calculated by said basic statistic parameter calculating means, whether the daily sale amount of the individual goods fit a Poisson distribution and, if the result is affirmative, for setting the daily sale amount as the Poisson type;

normal type distribution means for determining, by chi-square testing employing the basic statistic parameters calculated by said basic statistic parameter calculating means, whether the daily sale amount of the individual goods fit into a normal distribution and, if the result is affirmative, for setting the daily sale amount as the normal type;

causal type distribution means for determining whether coefficients of each causal in the linear regression formula is larger than a predetermined value and, if the result is affirmative, for setting the daily sale amount as the causal type; and

other type distribution means for setting the daily sale amount as the other type when the daily sale amount is not set to one of the Poisson type, the normal type, and the causal type.

10. The sale amount prediction system according to claim 4, wherein the sale amount predicting means sets the characteristic values as

mean value for the Poisson type,

mean value and standard deviation for the normal type,

linear multiple regression formula, mean value and standard deviation for the causal type, and

mean value and standard deviation for the other type.

15. The sale amount prediction system according to claim 4 wherein said basic statistic parameter calculating means calculates, based upon the sale amount data of individual goods calculated by said sale amount calculating means, a mean value, a standard deviation, a maximum value, a minimum value, a skewness value, a kurtosis value and a Geary value.

16. The sale amount prediction method according to claim 4, wherein the sale amount data of individual goods not consistently displayed and individual goods which are scrapped is removed from the sale amount data of the individual goods by said sale amount data calculating means, and the basic statistic parameters of the daily sale amount for the individual goods are calculated based on the daily sale amount data of the individual goods after such discarding.