Automatic selection of different visualizations for the organization of multivariate data

• US 9,892,530 B2
• Filed: 07/17/2012
• Issued: 02/13/2018
• Est. Priority Date: 06/07/2011
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Rope, Daniel J., Wills, Graham J.
• Primary Examiner(s): Le, Sarah
• Application Number:
  US13/550,985
• Publication Number:
  US 20120313949A1
• Time in Patent Office: 2,037 Days
• Field of Search: 345440
• CPC Class Codes: G06F 16/26 Visual data mining; Browsing structured data
• CPC Class Codes: G06F 16/9038 Presentation of query results
• CPC Class Codes: G06F 40/18 of spreadsheets form-filling G06F40/174
• CPC Class Codes: G06T 11/206 Drawing of charts or graphs
• CPC Class Codes: G16H 15/00 ICT specially adapted for medical reports, e.g. generation or transmission thereof

Abstract

According to one embodiment of the present invention, a computer-implemented method comprises generating a plurality of charts to visually represent a multivariate data set, mapping variables within the multivariate data set to components of each chart, calculating a score value for each chart based on a plurality of factors, and presenting one or more of the plurality of charts based on corresponding score value.

Claims

1. A computer-implemented method comprising:

generating a plurality of charts that visually represent data from a multivariate data set, wherein the generating a plurality of charts includes initially reducing a search space of potential chart candidates to be considered for presentation based on a corresponding score value by analyzing metadata associated with the multivariate data set in combination with statistical analysis of the data to determine relationships between variables of the multivariate data set, the determination of relationships between variables of the multivariate data set including a bivariate analysis that determines strengths for combinations of pairs of variables and a random forest model analysis that determines relationships between more than two variables;

mapping variables within the multivariate data set to components of each chart;

calculating a score value for each chart based on a plurality of factors; and

presenting one or more of the plurality of charts based on a corresponding score value for display, wherein the presenting further comprises facilitating navigation from a visual display of a first selected chart from the plurality of charts to a visual display of a second selected chart of the plurality of charts, wherein;

each variable associated with the first selected chart differs from each variable associated with the second selected chart, and

the facilitating of the navigation further comprises;

presenting a display having a plurality of nodes, each of the plurality of nodes representing a different respective variable within the multivariate data set, and each line connection between a pair of nodes representing a bivariate relationship between the respective different variables represented by the pair of nodes and being displayed differently based upon the bivariate relationship therebetween,

receiving a selection of an initial node, from among the plurality of nodes of the display, the initial node representing an initial variable, and

in response to the receiving of the selection, navigating to a subscreen having a visualization including the initial variable, for each visualization presenting at least one other variable such that more visualizations are known to exist that have meaningful statistical relationships, each selection of a variable from the presented at least one other variable adding a new variable and providing a different visual representation of the variables.

2. The method of claim 1, wherein one of the plurality of factors comprises strength of association between variables within the multivariate data set.

3. The method of claim 1, wherein one of the plurality of factors comprises suitability of each chart component and the variable mapped to that chart component.

4. The method of claim 1, wherein one of the plurality of factors comprises complexity of each chart.

5. The method of claim 1, wherein presenting one or more of the plurality of charts comprises presenting a subset of the plurality of charts, the subset of charts having greater score values.

6. A computer-implemented method comprising:

generating a plurality of charts that visually represent data from a multivariate data set;

mapping variables within the multivariate data set to components of each chart;

calculating a score value for each chart based on a plurality of factors; and

presenting one or more of the plurality of charts based on corresponding score value, the presenting further comprising facilitating navigation from a visual display of a first selected chart from the plurality of charts to a visual display of a second selected chart of the plurality of charts, wherein;

the facilitating of the navigation further comprises;

presenting a display having a plurality of nodes, each of the plurality of nodes representing a different respective variable in the multivariate data set, and each line connection between a pair of nodes representing a bivariate relationship between the respective different variables represented by the pair of nodes and being displayed differently based upon the bivariate relationship therebetween,

receiving a selection of an initial node, from among the plurality of nodes of the display, the initial node representing an initial variable, and

in response to the receiving of the selection, navigating to a subscreen having a visualization including the initial variable, for each visualization presenting at least one other variable such that more visualizations are known to exist that have meaningful statistical relationships, each selection of a variable from the presented at least one other variable adding a new variable and providing a different visual representation of the variables; and

the generating a plurality of charts includes initially reducing a search space of potential chart candidates to be considered for presentation based on a corresponding score value by analyzing metadata associated with the multivariate data set in combination with statistical analysis of the data to determine relationships between variables of the multivariate data set, the determination of relationships between variables of the multivariate data set including a bivariate analysis that determines strengths for combinations of pairs of variables and a random forest model analysis that determines relationships between more than two variables.

7. The method of claim 6, wherein one of the plurality of factors comprises strength of association between variables within the multivariate data set.

8. The method of claim 6, wherein one of the plurality of factors comprises suitability of each chart component and the variable mapped to that chart component.

9. The method of claim 6, wherein one of the plurality of factors comprises complexity of each chart.

10. The method of claim 6, wherein presenting one or more of the plurality of charts comprises presenting a subset of the plurality of charts, the subset of charts having greater score values.