Privacy Preserving Probabilistic Possibilistic Fuzzy C Means Clustering

Due to this uncontrollable growth of data, clustering played major role to partition into a small sets to do relevant processes within the small sets. Recently, the privacy and security are extra vital essentials when data is large and the data is distributed to other sources for various purposes. According to that, the privacy preservation should be done before distributing the data. In this study, our proposed algorithm meets the both requirements of achieving the clustering accuracy and privacy preserving of the data. Initially, the whole dataset is divided to small segments. The next step is to find the best sets of attributes combinations, which are attained through, attribute weighing process, which leads to attain the privacy preservation through vertical partitioning. The next is to apply the proposed Probabilistic Possibilistic Clustering Algorithm (PPFCM) for each segment, which produces the number of clusters for each segment. The next step is applying the PPFCM on the centroids of the clusters. The corresponding data tuples of the grouped centroids join to attain the final clustered result. The implementation is done using JAVA and the performance of the proposed PPFCM algorithm is compared with possibilistic FCM and probability-clustering algorithm for the benchmark datasets.