Performance Analysis of Hadoop Cluster for User Behavior Analysis

User activities produce an enormous amount of data when using popular devices such as smartphones. These data can be used to develop behavioral models in several areas including fraud detection, finance, recommendation systems, and marketing. However, enabling fast analysis of such a large volume of data using traditional data analytics tools may not be applicable. As a result, many organizations that are seeking to collect, process, and analyze big data have adopted a new class of technologies that includes Apache Hadoop and related tools. This paper reports on the feasibility and the performance of using a Hadoop cluster for user behavior analytics based on their activities in applications with a large number of users while using in-memory processing for faster querying and processing of data stored in computers memory rather than disk storage. In this paper, user behavior analysis is used as a base model for performance analysis because of its unique features and similarity with current problems in academia and industry, it includes performance analysis in two different areas: the performance of the cluster in data ingestion and its performance in analyzing the data.