DMCA
Value Decomposition and Dimension Selection in Multi-Dimensional Datasets using Map-Reduce Operation
BibTeX
@MISC{V_valuedecomposition,
author = {Preethi V and Suriya M},
title = {Value Decomposition and Dimension Selection in Multi-Dimensional Datasets using Map-Reduce Operation},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
The datasets which are in the form of object-attribute-time format is referred to as three-dimensional (3D) data sets. Clustering these three-dimensional (3D) data sets is a difficult task. So the subspace clustering method is applied to cluster the three-dimensional (3D) data sets. But finding the subspaces in the these three-dimensional (3D) dataset which is changing over time is really a difficult task. Sometimes this subspace clustering on three-dimensional (3D) data sets may produce the large number of arbitrary and spurious clusters. So to cluster these three-dimensional (3D) data sets a new centroid based concept is introduced called CATS. This CATS allows the users to select the preferred objects as centroids. This algorithm is not the parallel one. So it increases the time and space requirements which are needed to cluster the three-dimensional (3D) data sets. And in CATS no optimal centroids have been chosen to cluster the three-dimensional (3D) datasets. Since the CATS clusters the data based on the fixed centroids, the CATS cannot produce the good quality clusters. So for the first time in the proposed method the CPSO technique is introduced on the three-dimensional (3D) data sets to overcome all these drawbacks which clusters the three-dimensional (3D) datasets based on the optimal centroids and also it acts as the parallelization technique to tackle the space and time complexities.
Keyphrases
data set map-reduce operation multi-dimensional datasets dimension selection value decomposition optimal centroid difficult task cat cannot cpso technique large number object-attribute-time format new centroid parallelization technique fixed centroid parallel one good quality cluster called cat time complexity spurious cluster first time space requirement preferred object
