The grand tour: a tool for viewing multidimensional data
SIAM Journal on Scientific and Statistical Computing
Visual classification: an interactive approach to decision tree construction
KDD '99 Proceedings of the fifth ACM SIGKDD international conference on Knowledge discovery and data mining
Clustering through decision tree construction
Proceedings of the ninth international conference on Information and knowledge management
Machine Learning
ICDAR '95 Proceedings of the Third International Conference on Document Analysis and Recognition (Volume 1) - Volume 1
Interpretable Hierarchical Clustering by Constructing an Unsupervised Decision Tree
IEEE Transactions on Knowledge and Data Engineering
Toward Integrating Feature Selection Algorithms for Classification and Clustering
IEEE Transactions on Knowledge and Data Engineering
Targeted Projection Pursuit for Interactive Exploration of High- Dimensional Data Sets
IV '07 Proceedings of the 11th International Conference Information Visualization
A Projection Pursuit Algorithm for Exploratory Data Analysis
IEEE Transactions on Computers
Just-in-time annotation of clusters, outliers, and trends in point-based data visualizations
VAST '12 Proceedings of the 2012 IEEE Conference on Visual Analytics Science and Technology (VAST)
Visual pattern discovery using random projections
VAST '12 Proceedings of the 2012 IEEE Conference on Visual Analytics Science and Technology (VAST)
Hi-index | 0.00 |
Visualization technology makes it easier for users to spot patterns in data that would be difficult to find using only a computer algorithm. However, the discovery of a particular pattern is often only the first step in any analytical process, with the ultimate goal being insight into the underlying causes of this pattern. In current explorative interfaces, this analytical process often involves iterative hypothesis generation and testing, which gets exponentially more complex and time consuming as the dimensionality of the data set increases. In this paper, we suggest a technique that helps a user generate potential hypotheses for a particular observation or visual feature by reporting correlated dimensions. We use a modified decision tree algorithm that is not tuned for optimal classification, but for broad correlation detection. This paper presents the rationale for, algorithmic improvements in, and performance characteristics of the proposed technique, as well as a prototype implementation into a commercial data analysis tool.