Dimension Subset Selection for Data Visualization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computing systems face challenges in efficiently processing and visualizing large datasets, as they often present users with overwhelming numbers of dimensions, leading to reduced performance and user experience due to the need to rank and select relevant dimensions from a vast set.
Innovation Solution
A system that determines the cardinality and structural relationships of dimensions in a data model, filters out less relevant dimensions, and ranks others based on metrics, thereby reducing the search space and improving performance by providing a subset of dimensions for user selection, which enhances the visualization process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If all dimensions in a data model are presented to users for selection, then complete data exploration capability is provided, but system performance deteriorates and user experience worsens due to overwhelming numbers of dimensions
Solution Approach 1:
The system extracts and identifies a small subset of most relevant dimensions from the complete data model using cardinality analysis and structural relationship evaluation. By taking out only the essential dimensions (typically top 3-7) rather than presenting all dimensions, the system maintains data exploration capability while dramatically improving performance and user experience.
Solution Approach 2:
The dimension selection process is segmented into multiple analytical stages: calculating cardinalities, determining structural relationships, ranking dimensions based on multiple metrics, and filtering to produce a curated subset. This segmentation allows the system to process the complete dimension set efficiently while delivering a manageable subset to users.
2Adaptability or versatility
If all dimensions are presented for user selection, then comprehensive data analysis is enabled, but user experience deteriorates due to the need to rank and select from a vast set
Solution Approach 1:
The system extracts the most relevant dimensions based on cardinality and structural analysis, presenting only the essential subset (typically 3-7 dimensions) to users. This extraction eliminates the overwhelming experience of choosing from all dimensions while preserving the ability to perform comprehensive data analysis on the most impactful dimensions.
3Measurement precision
If cardinality calculation and structural relationship determination are performed for all dimensions, then accurate dimension ranking is achieved, but processing time and computational resources increase
Solution Approach 1:
The system performs preliminary calculations of dimension cardinalities and structural relationships once, storing these results for reuse. This preliminary action avoids repeated computation during dimension ranking operations, maintaining high ranking accuracy while significantly reducing processing time for subsequent queries.
Solution Approach 2:
The system calculates cardinalities and structural relationships for all dimensions (excessive action) but then uses these results to rank and select only the top subset. This approach ensures comprehensive and accurate analysis while limiting the final output to a manageable number of dimensions, balancing computational effort with practical utility.
Data Source
AI summary
Some embodiments provide a program that receives from a client device a dimension value for a first dimension in a plurality of dimensions of a data model and a measure in a set of measures of the data model. The program further determines a cardinality of dimension values for each dimension in the plurality of dimensions. The program also selects a subset of dimensions in the plurality of dimensions based on the cardinalities of dimension values for the plurality of dimensions. The program further sends the subset of dimensions in the plurality of dimensions to the client device. The program also receives from the client device a request for a chart visualization comprising data values associated with the measure and categorized according to a second dimension in the subset of dimensions. In response to the request, the program further sends the chart visualization to the client device.