Analyzing mining pattern evolutions using a data mining algorithm

• US 20070219990A1
• Filed: 03/16/2006
• Published: 09/20/2007
• Est. Priority Date: 03/16/2006
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: MacLennan, C., Cristofor, Elena, Crivat, Ioan
• Granted Patent: US 7,636,698 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/283 Multi-dimensional databases or data warehouses, e.g. MOLAP or ROLAP

Abstract

Architecture for analyzing pattern shifts in data patterns of data mining models and outputting the results. This allows comparing and describing differences between two semantically similar sets of patterns (or mining models), and for analyzing historical changes in versions of the same model or differences in patterns found by two or more different algorithms applied to the same data. The architecture can also facilitate explaining data patterns that shift over time and over different data populations, and between versions of the same model that use different algorithms. A model component is employed for storing data mining models have respective sets of data patterns obtained from a dataset, and an analysis component analyzes the sets of the data patterns for difference data therebetween. The dataset can be a subsample of a larger set of data and can be analyzed by the analysis component over a time period.

Claims

1. A computer-implemented system for analyzing mining pattern evolution, comprising:

a model component for storing data mining models have respective sets of data patterns obtained from a dataset; and

an analysis component for analyzing the sets of the data patterns for difference data therebetween.

2. The system of claim 1, wherein the difference data is obtained from sets of data patterns from a same dataset.

3. The system of claim 1, wherein the difference data is obtained from sets of data patterns from different datasets.

4. The system of claim 1, wherein the difference data is obtained from different versions of same data mining model, the mining models using different algorithms.

5. The system of claim 1, wherein the dataset is a subsample of a larger set of data.

6. The system of claim 1, wherein the difference data is analyzed by the analysis component over a time period.

7. The system of claim 1, further comprising a compare component that compares the data patterns.

8. The system of claim 1, further comprising an interface component for presenting the difference data as one of a decision tree, a dependency network, and a tabular listing of one or more rules.

9. The system of claim 1, further comprising a labeling component for uniquely labeling a data pattern.

10. The system of claim 9, wherein each of the data mining models include one of a classification algorithm and a clustering algorithm that facilitates labeling of the data pattern.

11. The system of claim 1, further comprising a machine learning and reasoning component that employs a probabilistic and/or statistical-based analysis to prognose or infer an action that a user desires to be automatically performed.

12. A computer-implemented method of processing data mining information for differences, comprising:

receiving a dataset;

processing some or all of the dataset for change information related to a change in the dataset using an algorithm of a data mining model; and

outputting the change information as one or more rules.

13. The method of claim 12, further comprising an act of outputting the rules to a user via a display.

14. The method of claim 12, further comprising an act of labeling data patterns of the dataset, and comparing the labels to determine the change information.

15. The method of claim 12, further comprising an act of labeling data patterns of the dataset, and comparing the data patterns to determine the change information.

16. The method of claim 12, further comprising an act of converting the change information into one or more rules.

17. The method of claim 12, further comprising an act of processing subsamples of the dataset with the algorithm to generate corresponding sets of data patterns, and comparing the sets of data patterns to obtain the change information.

18. The method of claim 12, further comprising acts of:

one of clustering and classifying data patterns of the dataset;

labeling the data patterns;

storing the labeled data patterns in the data mining model;

comparing the data patterns to generate the change information; and

updating the data mining model according to the one or more rules.

19. The method of claim 12, further comprising an act of processing all of the dataset with the algorithm at a first time and a second time, and comparing the change in the dataset from the first time to the second time to generate the change information.

20. A computer-executable system for processing data mining information, comprising:

computer-implemented means for receiving a dataset;

computer-implemented means for processing some or all of the dataset for change information related to a change in the dataset using an algorithm of a data mining model;

computer-implemented means for labeling data patterns of the dataset with corresponding labels;

computer-implemented means for comparing at least one of the labels and the data patterns for the change information; and

computer-implemented means for presenting the change information as one or more rules to a user.