A modified relationship based clustering framework for density based clustering and outlier filtering on high dimensional datasets

  • Authors:

  • Turgay Tugay Bilgin;A. Yilmaz Camurcu

  • Affiliations:

  • Department of Computer Engineering , Maltepe University, Maltepe, Istanbul, Turkey;Department of Electronics and Computer Education, Marmara University, Kadikoy, Istanbul, Turkey

  • Venue:
  • PAKDD'07 Proceedings of the 11th Pacific-Asia conference on Advances in knowledge discovery and data mining
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this study, we propose a modified version of relationship based clustering framework dealing with density based clustering and outlier detection in high dimensional datasets. Originally, relationship based clustering framework is based on METIS. Therefore, it has some drawbacks such as no outlier detection and difficulty of determining the number of clusters. We propose two improvements over the framework. First, we introduce a new space which consists of tiny partitions created by METIS, hence we call it micro-partition space. Second, we used DBSCAN for clustering micro-partition space. The visualization of the results are carried out by CLUSION. Our experiments have shown that, our proposed framework produces promising results on high dimensional datasets.