Clustering high dimensional data using RIA
Clustering may simply represent a convenient method for organizing a large data set so that it can easily be understood and information can efficiently be retrieved.However, identifying cluster in high dimensionality data sets is a difficult task because of the curse of dimensionality. Another chall...
Saved in:
| Main Author: | |
|---|---|
| Format: | Conference or Workshop Item |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://repo.uum.edu.my/18752/ http://doi.org/10.1063/1.4915706 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Clustering may simply represent a convenient method for organizing a large data set so that it can easily be understood and information can efficiently be retrieved.However, identifying cluster in high dimensionality data sets is a difficult task because of the curse of dimensionality. Another challenge in clustering is some traditional functions cannot capture the pattern dissimilarity among objects.In this article, we used an alternative dissimilarity measurement called Robust Influence Angle (RIA) in the partitioning method. RIA is developed using eigenstructure of the co variance matrix and robust principal component score. We notice that, it can obtain cluster easily and hence avoid the curse of dimensionality.It is also manage to cluster large data sets with mixed numeric and categorical value. |
|---|