Modified K-Means Algorithm for Big Data Clustering

摘要

Clustering of Big data is a highly demanding research issue and efficient clustering, particularly for growing data, attracts further attention to the researchers as it is a very common phenomenon for social networks. Clustering algorithms in general deal with static data and various algorithms do exist with their respective pros and cons and are applicable to various types of data. We consider K-means algorithm with one dimensional data and modify it to handle frequent addition of data without re-clustering the entire set. We further improve volume of distance matrix calculation for additional data elements. Theoretical calculation along with case study is placed for establishing the benefits gained by the proposed modified algorithm.