An Introduction to Clustering Algorithms in Big Data

Introduction

In Big Data, clustering is the process through which analysis is performed. Since the data is big, so it is very difficult to perform clustering approach. Big data is mainly termed as petabytes and zeta bytes of data and high computation cost is needed for the implementation of clusters. In this chapter we will show how clustering can be performed on Big Data and what are the different types of clustering approach. The challenge during clustering approach is to find observations within the time limit. Chapter also covers the possible future path for more advanced clustering algorithms. The chapter will cover single machine clustering and multiple machines clustering which also includes parallel clustering.

Today, data is increasing rapidly which actually forms a big data due to it high velocity, huge volume and different varieties of data (Shobana & Kumar, 2015). Few years before we are dealing with data collection challenges, but now in this era we are more concerned with processing this huge amount of data or big data (Nair & Bhagat, 2018a). Big data is analyzed for prediction and analysis purpose (Tsai, Lai, Chao, & Vasilakos, 2015). Scientists and researchers believe that today one of the most important topics in computer science is Big Data, whether the data collected through social networking websites, ecommerce websites or any other websites which is having accessibility to billions of users. You tube has more than 1 billion user which is producing almost 200 hours of video in each hour and its content ID service scans over 400 years of video every day. Main reason behind storing these types of data is further used for knowledge discovery (Chu, 2013). Data mining is the method which is mainly used for knowledge discovery and clustering is the process which is used for dividing the data into groups that contain objects of similar patterns (Chu, 2013). Clustering is vastly used in many areas such as bio-informatics (Y. Q. Zhang & Rajapakse, 2008), machine learning (Kubat, 2017), pattern recognition (Neal, 2007), networking (Sucasas et al., 2016) and lot of research work is done in this area. A decade before clustering algorithm was applied on small data in order to handle their complexity and computational cost, this also increase the speed with scalability. Due to occurrence of Big data in recent years more challenges are added to perform clustering.