Caviar Waterwheel Plant Optimization-Based Feature Selection For Early Stage Chronic Kidney Disease Big Data

Abstract

Chronic kidney disease, also termed as chronic renal disease or CKD, is a condition distinguished by a progressive loss in kidneys function over time. Diabetes as well as high blood pressure is the major frequent reason of CKD and sometimes it leads to death if it is left untreated. Several approaches were developed so far to find the CKD but, they failed due to lack of sufficient features and proper selection of features. In order to overcome such gaps, a hybrid optimization-driven model is developed for feature selection of early Stage of CKD utilizing Big data. The input big data is subjected to a data partitioning phase, which is conducted by Bayesian Fuzzy Clustering (BFC). Thereafter, the partitioned data is fed into the Spark architecture which includes two nodes

- a slave node and a master node. After that, pre-processing is carried out by the slave nodes utilizing the z-score normalization approach. Each node pre-processes a portion of the data and sends the output to the master node. The master node merges fused data from all the slave nodes. Finally, feature selection is done based on proposed Conditional Autoregressive Value-at-Risk Water Wheel Plant Algorithm (CWWPA), where fitness is based on entropy. The proposed CWWPA is devised by incorporating the Water Wheel Plant Algorithm (WWPA) and the CAViaR concept. CWWPA achieved maximum correlation coefficient of 0.948 also achieved minimum entropy of 0.399 with cluster size=6.