Predictive analytical modeling for databases

• US 8,443,013 B1
• Filed: 09/27/2011
• Issued: 05/14/2013
• Est. Priority Date: 07/29/2011
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Lin, Wei-Hao, Green, Travis H. K., Kaplow, Robert, Fu, Gang, Mann, Gideon S.
• Primary Examiner(s): Shmatov, Alexey
• Application Number:
  US13/246,541
• Time in Patent Office: 595 Days
• Field of Search: 707/810, 707/690, 707/999.103
• US Class Current: 707/810
• CPC Class Codes: G06F 16/2455 Query execution
• CPC Class Codes: G06N 20/00 Machine learning
• 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

Abstract

A computer-implemented method includes obtaining a database table, the table including multiple rows and multiple columns, in which one or more rows are missing at least one column value, executing a script, using a script engine, in response to obtaining the table, in which executing the script causes one or more values from the rows to be provided as input data to a first predictive model, and processing, using the first predictive model, the input data to obtain output data, the output data including a predicted value for at least one of the missing column values, and populating one or more of the missing column values with the output data to provide a revised database table.

Claims

1. A computer-implemented method comprising:

obtaining a database table, the table including a plurality of rows and a plurality of columns, wherein the database table has a plurality of missing column values;

executing a script, using a script engine, in response to obtaining the table, wherein executing the script causes a plurality of predictive models to be applied to the database table in a sequence order, and wherein the sequence order is determined according to a respective number of missing column values in the database table that each model can predict;

using each of the plurality of predictive models to generate respective output data, wherein the output data generated by each model includes a predicted value for at least one of the missing column values; and

populating the missing column values with the output data to provide a revised database table.

2. The computer-implemented method according to claim 1, wherein at least one of the predictive models comprises a customized predictive model.

3. The computer-implemented method according to claim 2, further comprising creating the customized predictive model based on a training dataset.

4. The computer-implemented method according to claim 3, wherein the training dataset includes the input data.

5. The computer-implemented method according to claim 3, wherein the training dataset includes data different from the input data.

6. The computer-implemented method of claim 2, wherein there are at least two missing column values in at least one of the table rows.

7. The computer-implemented method according to claim 2, further comprising selecting the customized predictive model from a plurality of different customized predictive models.

8. The computer-implemented method according to claim 7, further comprising training the plurality of customized predictive models, each predictive model being trained on a different arrangement of two or more missing column values in a particular table row.

9. The computer-implemented method of claim 1, wherein the sequence order is determined according to an increasing number of missing values that each model can predict.

10. The computer-implemented method of claim 9, wherein the database table comprises at least two columns containing missing values, and wherein executing the script causes a first predictive model configured to predict values for a corresponding column of the database table having the minimum number of missing values to be applied first in the sequence.

11. A system comprising:

one or more first computing devices configured to;

obtain a database table, the table including a plurality of rows and a plurality of columns, wherein the database table has a plurality of missing column values;

execute a script in response to obtaining the database table, wherein executing the script causes a plurality of predictive models to be applied to the database table in a sequence order, and wherein the sequence order is determined according to a respective number of missing column values in the database table that each model can predict;

process, using each of the plurality of predictive models to generate respective output data, wherein the output data generated by each model includes a predicted value for at least one of the missing column values; and

populate the missing column values with the output data to provide a revised database table.

12. The system according to claim 11, wherein at least one of the predictive models comprises a customized predictive model.

13. The system according to claim 12, wherein the one or more first computing devices are further configured to create the customized predictive model based on a training dataset.

14. The system according to claim 13, wherein the training dataset includes the input data.

15. The system according to claim 13, wherein the training dataset includes data different from the input data.

16. The system according to claim 12, wherein there are at least two missing column values in at least one of the table rows.

17. The system according to claim 12, wherein the one or more first computing devices are further configured to select the customized predictive model from a plurality of different customized predictive models.

18. The system according to claim 17, wherein the one or more first computing devices are further configured to train the plurality of customized predictive models, each predictive model being trained on a different arrangement of two or more missing column values in a particular table row.

19. A storage medium have instructions stored thereon that, when executed by data processing apparatus, cause the data processing apparatus to perform operations comprising:

obtaining a database table, the table including a plurality of rows and a plurality of columns, wherein the database table has a plurality of missing column values;

executing a script, using a script engine, in response to obtaining the table, wherein executing the script causes a plurality of predictive models to be applied to the database table in a sequence order, and wherein the sequence order is determined according to a respective number of missing column values in the database table that each model can predict;

processing, using each of the plurality of predictive models to generate respective output data, wherein the output data generated by each model includes a predicted value for at least one of the missing column values; and

populating the missing column values with the output data to provide a revised database table.

20. The storage medium according to claim 19, wherein at least one of the predictive models comprises a customized predictive model.

21. The storage medium according to claim 20, wherein the instructions, when executed by data processing apparatus, cause the data processing apparatus to perform further operations comprising creating the customized predictive model based on a training dataset.

22. The storage medium according to claim 21, wherein the training dataset includes the input data.

23. The storage medium according to claim 21, wherein the training dataset includes data different from the input data.

24. The storage medium according to claim 20, wherein there are at least two missing column values in at least one of the table rows.

25. The storage medium according to claim 20, wherein the instructions, when executed by data processing apparatus, cause the data processing apparatus to perform further operations comprising selecting the customized predictive model from a plurality of different customized predictive models.

26. The storage medium according to claim 25, wherein the instructions, when executed by data processing apparatus, cause the data processing apparatus to perform further operations comprising training the plurality of customized predictive models, each predictive model being trained on a different arrangement of two or more missing column values in a particular table row.