System and Method for Transaction Log Cleansing and Aggregation

• US 20090144122A1
• Filed: 11/30/2007
• Published: 06/04/2009
• Est. Priority Date: 11/30/2007
• Status: Active Grant

• Current Assignee: SAP SE
• Original Assignee: SAP AG (SAP SE)
• Inventors: Primozich, Neil, Ouimet, Kenneth, Warier, Prashant, Ginsberg, David
• Granted Patent: US 8,239,244 B2
• Time in Patent Office: Days
• US Class Current: 705/10
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities
• CPC Class Codes: G06Q 30/0204 Market segmentation

Abstract

A computer-implemented method prepares data for modeling. The method comprises storing data from customer sales transactions in a database and retrieving a dataset of the data from the database. The dataset may include promotion and merchandizing entries. The method includes cleansing the dataset to remove erroneous and anomalous entries. Cleansing the dataset may include determining a threshold value from the dataset and determining whether a value of the dataset exceeds the threshold value, and determining an out-of-stock status for a product from the dataset. The method includes aggregating the dataset over a plurality of dimensions of the transactional space including store, product, and time dimensions, and analyzing the dataset following the cleansing and aggregating steps within a model to predict attributes of subsequent sales transactions.

Claims

1. A computer-implemented method of preparing data for modeling, comprising:

storing data from customer sales transactions in a database;

retrieving a dataset of the data from the database;

cleansing the dataset to remove erroneous and anomalous entries;

aggregating the dataset over a plurality of dimensions of the transactional space including store, product, and time dimensions; and

analyzing the dataset following the cleansing and aggregating steps within a model to predict attributes of subsequent sales transactions.

2. The computer implemented method of claim 1, wherein cleansing the dataset includes determining an out-of-stock status of a product from the dataset.

3. The computer implemented method of claim 2, wherein determining an out-of-stock status of a product from the dataset includes:

determining a probable number of sales of a product from the dataset;

determining an actual number of sales of a product from the dataset; and

comparing the actual number of sales of a product to the probable number of sales of a product.

4. The computer implemented method of claim 1, wherein the model includes a function of a number of customers who had access to a product at a point within the transactional space and sales per customer at the point within the transactional space.

5. The computer implemented method of claim 1, wherein aggregating the dataset involves summing entries from the dataset within the same dimensional space.

6. The computer implemented method of claim 1, wherein the dataset further includes promotion and merchandizing entries.

7. The computer implemented method of claim 1, wherein cleansing the dataset includes:

determining a threshold value from the dataset; and

determining whether a value of the dataset exceeds the threshold value.

8. The computer implemented method of claim 1, including:

determining a historical number of customers for a given store and a given time from the dataset; and

forecasting a future number of customers for a given store and a given time from the historical number of customers for a given store and a given time using exponential smoothing techniques.

9. The computer implemented method of claim 1, wherein analyzing the dataset includes predicting sales of a product given a price, promotion and merchandizing.

10. The computer implemented method of claim 1, wherein analyzing the dataset includes determining an optimal price for a product.

11. A method of preparing data for modeling, comprising:

retrieving a dataset from a database, wherein the dataset includes data from customer sales transactions;

cleansing the dataset to remove erroneous and anomalous entries;

aggregating the dataset over a plurality of dimensions of the transactional space including store, product, and time dimensions; and

transmitting the dataset to a modeling system, wherein the modeling system predicts attributes of subsequent sales transactions.

12. The method of claim 11, wherein cleansing the dataset includes determining an out-of-stock status of a product from the dataset.

13. The method of claim 12, wherein determining an out-of-stock status of a product from the dataset includes:

determining a probable number of sales of a product from the dataset;

determining an actual number of sales of a product from the dataset; and

comparing the actual number of sales of a product to the probable number of sales of a product.

14. The method of claim 11, wherein the modeling system includes a function of a number of customers who had access to a product at a point within the transactional space and sales per customer at the point within the transactional space.

15. The method of claim 11, wherein cleansing the dataset includes:

determining a threshold value from the dataset; and

determining whether a value of the dataset exceeds the threshold value.

16. The method of claim 11, wherein the modeling system predicts sales of a product given a price, promotion and merchandizing.

17. The method of claim 11, wherein the modeling system determines an optimal price for a product.

18. A computer program product usable with a programmable computer processor having a computer readable program code embodied therein, comprising:

computer readable program code which stores data from customer sales transactions in a database;

computer readable program code which retrieves a dataset of the data from the database;

computer readable program code which cleanses the dataset to remove erroneous and anomalous entries;

computer readable program code which aggregates the dataset over a plurality of dimensions of the transactional space including store, product, and time dimensions; and

computer readable program code which analyzes the dataset within a model to predict attributes of subsequent sales transactions.

19. The computer program product of claim 18, wherein the computer readable program code which cleanses the dataset to remove erroneous and anomalous entries includes computer readable program code which determines an out-of-stock status of a product from the dataset.

20. The computer program product of claim 19, wherein the computer readable program code which determines an out-of-stock status of a product from the dataset includes:

computer readable program code which determines a probable number of sales of a product from the dataset;

computer readable program code which determines an actual number of sales of a product from the dataset; and

computer readable program code which compares the actual number of sales of a product to the probable number of sales of a product.

21. The computer program product of claim 18, wherein the computer readable program code which analyzes the dataset within a model includes computer readable program code which determines an optimal price for a product.

22. A computer system for manipulating data, comprising:

means for retrieving a dataset from a database, wherein the dataset includes data from customer sales transactions;

means for cleansing the dataset to remove erroneous and anomalous entries;

means for aggregating the dataset over a plurality of dimensions of the transactional space including store, product, and time dimensions; and

means for transmitting the dataset to a modeling system, wherein the modeling system predicts attributes of subsequent sales transactions.

23. The computer system of claim 22, wherein the means for cleansing the dataset includes:

means for determining a threshold value from the dataset; and

means for determining whether a value of the dataset exceeds the threshold value.

24. The computer system of claim 22, wherein the means for cleansing the dataset includes means for determining an out-of-stock status of a product from the dataset.

25. The computer system of claim 24, wherein the means for determining an out-of-stock status of a product from the dataset includes:

means for determining a probable number of sales of a product from the dataset;

means for determining an actual number of sales of a product from the dataset; and

means for comparing the actual number of sales of a product to the probable number of sales of a product.