HFMST: An Efficient Adaptive Fuzzy Linkage Feature Selection with Hybrid Fuzzy Based Minimum Spanning Tree (HFMST) Clustering Algorithm

Abstract

In this paper, the HFMST algorithm—a novel hybrid Fuzzy Based Minimum Spanning Tree (HFMST) clustering—is presented. A new clustering approach known as HFMST clustering algorithm is created by merging Enhanced Adaptive Fuzzy Linkage based Feature Selection (EAFFS) algorithm with MST. The current feature selection approaches have a high likelihood of redundant features showing up in the final subset, but in most circumstances, locating and eliminating them can significantly increase the clustering accuracy. EAFFS and the HFMST methodology are two approaches used to address this issue. The standard clustering centre is upgraded to take into account the sample weight and the cluster centre of the cluster group when determining the position in the cluster tree. Up to the point of the algorithm convergence iteration optimizes the cluster centres and partitions while also using MST to recalculate the cluster centre. Comparing the proposed HFMST algorithm to other clustering algorithms like DBSCAN, DCNaN, and RDMN algorithms reveals that it has a high Rand Index (RI) and Adjusted Rand Index (ARI) ratio.