Linear Interpolation-Based Fuzzy Clustering Approach for Missing Data Handling

Abstract

Clustering of incomplete data set containing missing values is a common problem in the literature. Methods to handle this problem have vast variations, including several imputation as well as non-imputation techniques for clustering. In this work, we have described the analysis of different approaches explored for handling missing data in clustering. The aim of this paper is to compare several FCM clustering approaches based on imputation and non-imputation strategies. Experimental results on one artificial and four real-world data sets from UCI repository show that linear interpolation-based FCM clustering approach performs significantly better than other techniques for these data sets.