Adaptive Threshold Based Clustering: A Deterministic Partitioning Approach

Adaptive Threshold Based Clustering: A Deterministic Partitioning Approach

Mamta Mittal (Department of Computer Science and Engineering, G. B. Pant Govt. Engineering College, New Delhi, India), Rajendra Kumar Sharma (Thapar University, Patiala, India), Varinder Pal Singh (Thapar University, Patiala, India) and Raghvendra Kumar (Department of Computer Science and Engineering, LNCT College, Indore, India)
Copyright: © 2019 |Pages: 18
DOI: 10.4018/IJISMD.2019010103

Abstract

Partitioning-based clustering methods have various challenges especially user-defined parameters and sensitivity to initial seed selections. K-means is most popular partitioning based method while it is sensitive to outlier, generate non-overlap cluster and non-deterministic in nature due to its sensitivity to initial seed selection. These limitations are regarded as promising research directions. In this study, a deterministic approach which do not requires user defined parameters during clustering; can generate overlapped and non-overlapped clusters and detect outliers has been proposed. Here, a minimum support value has been adopted from association rule mining to improve the clustering results. Further, the improved approach has been analysed on artificial and real datasets. The results demonstrated that datasets are well clustered with this approach too and it achieved success to generate almost same number of clusters as present in real datasets.
Article Preview

2. Literature Review

In 1960s, analysts utilized “Data Fishing” or Data Digging” for getting information from the repositories. They developed numerous methods to parcel it into various groups where objects in each group is similar to one another. For this purpose, MacQueen (1967) developed k-means method for acquiring k clusters, where each cluster is separated from other cluster based on the distance from the centroid and data objects. From that point onward, Salton and Wong (1971) acquainted single pass clustering for parcel it into disjoint sets. Duda and Hart (1973) portrayed how parceling the datasets is important in pattern recognition and acknowledgment. Dubes and Jain (1976) emphasized that the user should be aware of intrinsic properties of a clustering technique before using them. Lloyd (1982) presented k-means method for apportioning the datasets. In this, separation between the data objects and centroids is least as they are at the minimum distance. On the other hand, in 1987, Kaufman presented k-medoid method, depends on the hunt of k delegates called medoids for the provided databases.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 10: 4 Issues (2019): 1 Released, 3 Forthcoming
Volume 9: 4 Issues (2018)
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing