Predictive analytical model selection

• US 8,694,540 B1
• Filed: 09/27/2011
• Issued: 04/08/2014
• Est. Priority Date: 09/01/2011
• 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): SYED, FARHAN M
• Application Number:
  US13/246,410
• Time in Patent Office: 924 Days
• US Class Current: 707/777
• CPC Class Codes: G06F 16/211 Schema design and management

Abstract

A computer-implemented method includes obtaining a database table, the database table including data arranged in a plurality of rows and a plurality of columns, each column of data being associated with a different tag that specifies a category for data in the column, using one or more processors to identify a first predictive model, from a collection of predictive models, that can be applied to the database table to generate a predictive output, in which identifying the first predictive model is based on one or more of the different tags, adding a name associated with the first predictive model to a set of names of predictive models that are compatible with the database table, and providing the set of names of predictive models to a client device.

Claims

1. A computer-implemented method, the method comprising:

obtaining, over a network, a database table from a client device, the database table including data arranged in a plurality of rows and a plurality of columns, each column of data being associated with a different tag that specifies a category for data in the column, the different tag separate from a column name associated with each column of data;

receiving a query to identify at least one predictive model compatible with the database table;

in response to receiving the query, using one or more processors to identify, based on one or more of the different tags, a plurality of predictive models, from a collection of predictive models, that can be applied to the database table to generate a predictive output, at least one identified predictive model being trained on data selected from a group consisting of data associated with one or more categories related to categories specified by the different tags, data associated with the same categories specified by the different tags and arranged in a different order from the data of the database table, and combinations thereof;

ranking the identified predictive models based on how closely categories of data accepted by each identified predictive model match the categories of data specified by the different tags, wherein the categories of data accepted by each identified predictive model are categories of data on which the identified predictive model has been trained;

adding a name associated with at least one of the identified predictive models to a set of names of predictive models that are compatible with the database table; and

providing the set of names of predictive models to the client device.

2. The computer-implemented method of claim 1, further comprising:

applying at least one of the identified predictive models to the database table to obtain one or more predicted values; and

adding the one or more predicted values to the database table.

3. The computer-implemented method of claim 2, wherein adding the one or more predicted values to the database table comprises replacing a missing column value in the database table.

4. The computer-implemented method of claim 2, wherein adding the one or more predicted values to the database table comprises adding a new column to the database table, the new column including at least one predicted value.

5. The computer-implemented method of claim 1, wherein identifying the plurality of predictive models comprises identifying a first predictive model configured to:

receive, as an input, data in a category specified by at least one of the different tags; and

output data in a category specified by a second tag.

6. The computer-implemented method of claim 5, wherein identifying the plurality of predictive models further comprises identifying

a second predictive model from the collection of predictive models, the second predictive model being configured to receive, as an input, both data in a category specified by at least one of the different tags and data in a category specified by the second tag;

and

adding a name associated with the second predictive model to the set of names of predictive models that are compatible with the database table.

7. The computer-implemented method of claim 6, further comprising:

applying the first predictive model to the database table to obtain a first predicted value;

adding the first predicted value to the database table in a new column to obtain an updated database table;

subsequently applying the second predictive model to the updated database table to obtain a second predicted value; and

adding the second predicted value to the updated database table.

8. The computer-implemented method of claim 1, wherein identifying the plurality of predictive models comprises identifying a predictive model configured to receive, as an input, data in a category specified by a second tag related to at least one of the different tags.

9. The computer-implemented method of claim 8, wherein the second tag is superior to or subordinate to the at least one of the different tags in a hierarchal order of tags.

28. The computer-implemented method of claim 1, wherein the tags are arranged in a first order, and wherein using the one or more processors to identify the plurality of predictive models from the collection of predictive models comprises:

arranging the tags in a second order; and

using the one or more processors to identify a predictive model that takes, as an input, categories of data arranged according to the second order.

29. The computer-implemented method of claim 1, wherein the predictive models in the collection of predictive models were trained using training data provided by a computing system other than the client device.

30. The computer-implemented method of claim 1, wherein ranking the identified predictive models comprises:

(a) determining, for each identified predictive model, a number of categories of data accepted by the identified predictive model that are related to or exactly match categories of data specified by the different tags; and

(b) ranking the identified predictive models in order according to the number determined in (a).

31. The computer-implemented method of claim 1, wherein the one or more categories related to categories specified by the different tags are in a same category hierarchy as the categories specified by the different tags.

10. A system comprising:

one or more computing devices configured to perform operations including;

obtaining, over a network, a database table from a client device, the database table including data arranged in a plurality of rows and a plurality of columns, each column of data being associated with a different tag that specifies a category for data in the column, the different tag separate from a column name associated with each column of data;

receiving a query to identify at least one predictive model compatible with the database table;

in response to receiving the query, using one or more processors to identify, based on one or more of the different tags, a plurality of predictive models, from a collection of predictive models, that can be applied to the database table to generate a predictive output, at least one identified predictive model being trained on data selected from a group consisting of data associated with one or more categories related to categories specified by the different tags, data associated with the same categories specified by the different tags and arranged in a different order from the data of the database table, and combinations thereof;

ranking the identified predictive models based on how closely categories of data accepted by each identified predictive model match the categories of data specified by the different tags, wherein the categories of data accepted by each identified predictive model are categories of data on which the identified predictive model has been trained;

adding a name associated with at least one of the identified predictive models to a set of names of predictive models that are compatible with the database table; and

providing the set of names of predictive models to the client device.

11. The system of claim 10, wherein the one or more computing devices are further configured to perform operations including:

applying at least one of the identified predictive models to the database table to obtain one or more predicted values; and

adding the one or more predicted values to the database table.

12. The system of claim 11, wherein adding the one or more predicted values to the database table comprises replacing a missing column value in the database table with at least one predicted value.

13. The system of claim 11, wherein adding the one or more predicted values to the database table comprises adding a new column to the database table, the new column including at least one predicted value.

14. The system of claim 10, wherein identifying the plurality of predictive models comprises identifying a first predictive model configured to:

receive, as an input, data in a category specified by at least one of the different tags; and

output data in a category specified by a second tag.

15. The system of claim 14,

wherein identifying the plurality of predictive models further comprises identifying a second predictive model from the collection of predictive models, the second predictive model being configured to receive, as an input, both data in a category specified by at least one of the different tags and data in a category specified by the second tag;

and

adding a name associated with the second predictive model to the set of names of predictive models that are compatible with the database table.

16. The system of claim 15, wherein the one or more computing devices are further configured to perform operations including:

applying the first predictive model to the database table to obtain a first predicted value;

adding the first predicted value to the database table in a new column to obtain an updated database table;

subsequently applying the second predictive model to the updated database table to obtain a second predicted value; and

adding the second predicted value to the updated database table.

17. The system of claim 10, wherein identifying the plurality of predictive models comprises identifying a predictive model configured to receive, as an input, data in a category specified by a second tag related to at least one of the different tags.

18. The system of claim 17, wherein the second tag is superior to or subordinate to the at least one of the different tags in a hierarchal order of tags.

32. The system of claim 10, wherein the one or more categories related to categories specified by the different tags are in a same category hierarchy as the categories specified by the different tags.

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

obtaining, over a network, a database table from a client device, the database table including data arranged in a plurality of rows and a plurality of columns, each column of data being associated with a different tag that specifies a category for data in the column, the different tag separate from a column name associated with each column of data;

receiving a query to identify at least one predictive model compatible with the database table;

in response to receiving the query, using one or more processors to identify, based on one or more of the different tags, a plurality of predictive models, from a collection of predictive models, that can be applied to the database table to generate a predictive output, at least one identified predictive model being trained on data selected from a group consisting of data associated with one or more categories related to categories specified by the different tags, data associated with the same categories specified by the different tags and arranged in a different order from the data of the database table, and combinations thereof;

ranking the identified predictive models based on how closely categories of data accepted by each identified predictive model match the categories of data specified by the different tags, wherein the categories of data accepted by each identified predictive model are categories of data on which the identified predictive model has been trained;

adding a name associated with at least one of the identified predictive models to a set of names of predictive models that are compatible with the database table; and

providing the set of names of predictive models to the client device.

20. The storage medium according to claim 19 wherein the instructions, when executed by data processing apparatus, cause the data processing apparatus to perform further operations comprising:

applying at least one of the identified predictive models to the database table to obtain one or more predicted values; and

adding the one or more predicted values to the database table.

21. The storage medium according to claim 20, wherein adding the one or more predicted values to the database table comprises replacing a missing column value in the database table with the predicted value.

22. The storage medium according to claim 20, wherein adding the one or more predicted values to the database table comprises adding a new column to the database table, the new column including at least one predicted value.

23. The storage medium according to claim 19, wherein identifying the plurality of predictive models comprises identifying a first predictive model configured to:

receive, as an input, data in a category specified by at least one of the different tags; and

output data in a category specified by a second tag.

24. The storage medium according to claim 23, wherein identifying the plurality of predictive models further comprises identifying a second predictive model from the collection of predictive models, the second predictive model being configured to receive, as an input, both data in a category specified by at least one of the different tags and data in a category specified by the second tag;

and

adding a name associated with the second predictive model to the set of names of predictive models that are compatible with the database table.

25. The storage medium according to claim 24, wherein the instructions, when executed by data processing apparatus, cause the data processing apparatus to perform further operations comprising:

applying the first predictive model to the database table to obtain a first predicted value;

adding the first predicted value to the database table in a new column to obtain an updated database table;

subsequently applying the second predictive model to the updated database table to obtain a second predicted value; and

adding the second predicted value to the updated database table.

26. The storage medium according to claim 19, wherein identifying the plurality of predictive models comprises identifying a predictive model configured to receive, as an input, data in a category specified by a second tag related to at least one of the different tags.

27. The storage medium according to claim 26, wherein the second tag is superior to or subordinate to the at least one of the different tags in a hierarchal order of tags.

33. The storage medium according to claim 20, wherein the one or more categories related to categories specified by the different tags are in a same category hierarchy as the categories specified by the different tags.