A comparative study to classify big data using fuzzy techniques

Abstract

It is very difficult to implement an efficient analysis by using the customary techniques currently available; this is due to the fact that the data size has had a huge increase. Many complications were faced because of the numerous characteristics of big data; some of them include complexity, value, variability, variety, velocity, and volume. The objective of this paper is to implement classification techniques using the map reduce framework using fuzzy and crisp methods, also to arrange for a study that can compare and contrast the outcomes of the suggested systems against the methods appraised in the documented works. For this research the applied method for the fuzzy technique is the fuzzy k-nearest neighbor, and for the non-fuzzy techniques both the support vector machine and the k-nearest neighbor are used. The use of the map reduce paradigm is applied to be able to process big data. We also implemented an integrated system using the Support Vector Machine with the fuzzy soft label and Gaussian fuzzy membership. Results show that fuzzy k-nearest neighbor classifier gives higher accuracy but it takes a lot of time in classification compared to the other techniques. But the outcomes when projected onto other data sets demonstrate that the suggested method that used fuzzy logic in the Reducer function gives higher accuracy and lower time than the new suggested methods and the methods revised in the paper.