Multi-objective Hybrid Fuzzified PSO and Fuzzy C-Means Algorithm for Clustering CDR Data

Abstract

The growing field of mobile telecommunication becomes more and more competitive in the world. Therefore, the mobile operators are facing tremendous challenges. The customer requirements have increased in various levels of environments with huge amount of complexity of data. To solve this complexity of the problem a single-objective function may not offer suitable results of complex data. So this paper proposes a novel fuzzy clustering method based on Multi-Objective Hybrid Fuzzified particle swarm optimization and Fuzzy C-Means algorithm for the complex data. In this multi-objective hybrid clustering algorithm uses two objective functions: JFCM and the Xie-Beni index function. In the optimization level, FPSO find an optimal number of clusters by minimizing the value by Xie-Beni index. Based on the obtained number of clusters, FCM utilizes the optimal clustering centers and output the clustering results using JFCM objective function. Fuzzy C-Means (FCM) algorithm is the most liked and widely used clustering technique, because it is efficient and easy to execute. However, it suffers from the problem of initialization and is easily trapped in local optima. This clustering technique is used for predicting the social behavior and people's common interest, anti social thread by anti social communication. Then, based on business requirement, the service provider can predict which tower has to be changed, where to build a new tower and other such details based on the clustered data of customer's activities and behavior of different time period. The experimental results are shown for mobile call data records to obtain optimum and fast clustering solution.