A hybrid clustering technique to improve big data accessibility based on machine learning approaches

Abstract

Big data is called to a large or complex data from traditional ones, which is unstructured in many case. Accessing to a specific value in a huge data that is not sorted or organized can be time consuming and require a high processing. With growing of data, clustering can be a most important unsupervised approach that finds a structure for data. In this paper, we demonstrate two approaches to cluster data with high accuracy, and then we sort data by implementing merge sort algorithm finally, we use binary search to find a data value point in a specific range of data. This research presents a high value efficiency combo method in big data by using genetic and k-means. After clustering with k-means total sum of the Euclidean distances is 3.37233e+09 for 4 clusters, and after genetic algorithm this number reduce to 0.0300344 in the best fit. In the second and third stage we show that after this implementation, we can access to a particular data much faster and accurate than other older methods.