Goal seeking using predictive analytics

• US 7,788,200 B2
• Filed: 02/02/2007
• Issued: 08/31/2010
• Est. Priority Date: 02/02/2007
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Crivat, Ioan Bogdan, MacLennan, C. James, Iyer, Raman S.
• Primary Examiner(s): Sparks; Donald
• Assistant Examiner(s): Kim; David H
• Application Number:
  US11/670,656
• Publication Number:
  US 20080189237A1
• Time in Patent Office: 1,306 Days
• Field of Search: 707/2, 706/46, 715/212
• US Class Current: 706/46
• CPC Class Codes: G06N 7/00 Computing arrangements based on specific mathematical models

Abstract

Seeking goals in data that can be expressed as rows and columns is provided through predictive analytics. If a desired goal is achievable, the changes to the rows and/or columns that can achieve the goal are presented to a user. If the desired goal is not achievable, an error message or other indicator can be presented to the user. Predictive analytics can include a predictive algorithm, various data mining techniques, or other predictive techniques. A confidence metric of a goal-seek result can be normalized to estimate the degree of confidence that a particular change will yield the desired outcome.

Claims

1. A system that facilitates goal seeking in tabular data, comprising:

a processor;

system memory;

a receiver component that receives a desired goal based on tabular data, the goal comprising a substantially increased likelihood that the tabular data of a target column has a target value, wherein the receiver component is configured to receive and process both categorical goals comprising at least a target column that contains a finite set of categorical values and continuous goals comprising the target column that contains values in a range or interval;

an optimization component that optimizes goal seeking for at least one degree of freedom, wherein optimizing includes determining patterns for a selected row that determine a value of a target column as a function of the other columns within the tabular data, the determined patterns providing at least one row value that increases the likelihood of reaching the desired goal in the target column, the row values being inferred using probabilistic inference, wherein the probabilistic inference includes computing a probability distribution over states of interest based on a consideration of data and events of rows other than the selected row, wherein the row values are based on the probability they yield for the goal value if categorical goals were received and are based on the closest estimation to the goal if continuous goals were received; and

an output component that presents recommended actions within the tabular data to achieve the desired goal.

2. The system of claim 1, further comprising a measurement component that measures a degree of confidence that the recommended action will yield the desired goal.

3. The system of claim 2, the measurement component measures the degree of confidence directly from a predictive model generated by the optimization component.

4. The system of claim 1, the optimization component optimizes goal seeking utilizing at least one of a classification model, or an estimation model, or both.

5. The system of claim 1, the optimization component optimizes goal seeking for a plurality of degrees of freedoms utilizing a greedy approach.

6. The system of claim 1, the output component presents an error message if the optimization component can not optimize the goal seeking.

7. The system of claim 1, the optimization component describes the target column values with the function:

F=(Col1Values×

Col2Values×

. . . ×

ColN values)→

(targetColValues).

8. The system of claim 1, the optimization component further identifies fixed factors that should not be optimized.

9. The system of claim 1, the optimization component is executed on a server independent of the location of the tabular data.

10. The system of claim 1, the tabular data is contained in a spreadsheet application.

11. The system of claim 1, the tabular data is contained in an XML application.

12. A computer-implemented dynamic goal method for dynamic goal seeking in tabular data, comprising:

receiving an input that includes a desired goal expressed within tabular data, the goal comprising a substantially increased likelihood that the tabular data of a target column has a target value, wherein receiving includes receiving and processing both categorical goals comprising at least a target column that contains a finite set of categorical values and continuous goals comprising the target column that contains values in a range or interval;

uploading the tabular data as at least one rowset parameter; and

building a predictive model based in part on the at least one rowset parameter and the desired goal, wherein the desired goal is optimized, the optimizing includes determining patterns for a selected row that determine a value of a target column as a function of the other columns within the tabular data, the determined patterns providing a row value that increases the likelihood of reaching the desired goal in the target column, the row values being inferred using probabilistic inference, wherein the probabilistic inference includes computing a probability distribution over states of interest based on a consideration of data and events of rows other than the selected row, wherein the row values are based on the probability they yield for the goal value if categorical goals were received and are based on the closest estimation to the goal if continuous goals were received.

13. The method of claim 12, building a predictive model comprising utilizing a classification model or an estimation model.

14. The method of claim 12, further comprising:

ascertaining at least one value for each column contained in the tabular data;

exploring possible combinations for each column; and

determining if the desired goal is achievable.

15. The method of claim 14, further comprising returning at least one combination that maximizes the probability if the desired goal is achievable.

16. The method of claim 14, further comprising sending an error message if the desired goal is not achievable.

17. The method of claim 12, further comprising identifying fixed factors and eliminating the fixed factors from the built predictive model.

18. A computer executable system including a processor and system memory that provides recommends actions to achieve a goal relating to tabular data, comprising:

means for receiving a desired goal represented by a plurality of columns and a plurality of rows, the goal comprising a substantially increased likelihood that the tabular data of a target column has a target value, wherein receiving includes receiving and processing both categorical goals comprising at least a target column that contains a finite set of categorical values and continuous goals comprising the target column that contains values in a range or interval;

means for evaluating the plurality of columns and the plurality of rows, the evaluating including optimizing the desired goal, wherein the optimizing includes determining patterns for a selected row that determine a value of a target column as a function of the other columns within the tabular data, the determined patterns providing a row value that increases the likelihood of reaching the desired goal in the target column, the row values being inferred using probabilistic inference, wherein the probabilistic inference includes computing a probability distribution over states of interest based on a consideration of data and events of rows other than the selected row, wherein the row values are based on the probability they yield for the goal if continuous goals were received;

means for determining a change to at least one of the plurality of columns or at least one of the plurality of row to achieved the desired result or both; and

means for presenting the results of the determination.

19. The computer executable system of claim 18, further comprising means for determining a degree of confidence that the determined change will achieve the desired goal.

20. The computer executable system of claim 18, further comprising means for building at least one of a classification model or an estimation model or both the classification model and estimation model.