Investigation of computational Algorithms for enhancing data Clustering techniques using feature Linkage weight based feature Reduction

Abstract

Emergence of modern technologies has led to huge volumes of high dimensional data. Technologies like IoT, Cloud, and data analytics demand measuring, storing, and analysing huge volumes of data with extremely high number of features. There are several challenges associated with increase in dimension of data and one major challenge is the analysis of high-dimensional data. High dimensional data faces the problem of dimensionality curse, where only small search space is used for the analysis of the problem instead of the whole problem space. As the solution of the problem is analysed only on the subset of the search space, this may allow to descend the finding of the global optimum solution. newlineData mining is the process of extraction of patterns from massive volume of data, where mining involves the use of data analysis tools to realize previously unidentified patterns and relationships from huge amount of data. Presently, enormous growth of information increases the mining challenge of investigating the huge volume of data in extracting the patterns. To eradicate the complexity of handling huge data volume and adopting the data to make it suitable for analysis, reduction in dimension must be performed using feature reduction techniques. newlineThe main objective of the research work is to overcome the challenge of high-dimensional data by reducing the number of features in the dataset and to improve both the computational time and cluster formation accuracy. The first contribution of the research work proposes an algorithm FRFCM-FLW to undergo feature reduction in the Fuzzy C-Means clustering algorithm using the feature linkage weight criteria. newline