An efficient clustering mechanism in big data framework for data preprocessing and management

Abstract

An efficient data handling mechanism has been applied based on epoch-based k-means associated fuzzy clustering (EKFC). In the first phase weights have been assigned to individual data segment presented based on the classification key metrics. It has been assigned automatically. Then weight preprocessing has been done in such manner to prune the unwanted weights. It has been pruned in such way to filter the weights which are not scalable. Then epoch-based k-means associated fuzzy clustering (EKFC) approach has been applied for data arrangement. First different epochs have been considered for the calculation of initial seeds values. These seeds have been considered after considering 100 epochs. After 100 epochs seeds have been determined. These seeds values have been used as the initial centroid for the k-means clustering. After the complete validation similar clusters from the two clustering approaches have been considered. In the next phase operational clustering has been performed. In the final phase threshold ranking has been performed. It has been performed for the final classification based on the above clusters. It will arrange in the order of threshold values. It will be used for the determination of the priority of the task in the big data environment. The results are found to be prominent in terms of classification accuracy.