Deep Fuzzy Clustering with Political Optimization Enabled Centroid Update Model for Incremental Data Clustering in Mapreduce Framework

The big data arrives continuously by different modes like social networks, e-commerce, and online shopping platforms. The major issue in the development of contemporary clustering technique is that in several tools, new data are added into large databases and thus it becomes impractical to perform data clustering. Thus, this research study proposes a novel optimization-driven technique for incremental big data clustering by the Mapreduce model. The input data are initially provided to the mapper for feature selection, which is done using Tanimoto similarity. The clustering is achieved in reducer with Deep Fractional Calculus Political Optimizer fuzzy clustering (Deep FC-PO fuzzy clustering). The proposed FC-PO adapted for tuning the attributes of deep fuzzy clustering (DFC). Here, the FC-PO is formed by combining Fractional calculus (FC), and Political Optimizer (PO). Moreover, incremental clustering is done by matching incremental data with the centroid using the optimized Tversky index, wherein the coefficients of Tversky index is evaluated by FC-PO. After incremental clustering, the centroid update is done based on Deep Residual Network (DRN). The FC-PO achieved the high clustering accuracy of 90.6%, greater jaccard coefficient of 0.889 and highest random coefficient of 0.986.