User Behavior Analysis using Smartphones

Abstract

Users' 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 may not be applicable. In-memory analytics is a new technology for faster querying and processing of data stored in computer's memory (RAM) rather than disk storage. This research reports on the feasibility of user behavior analytics based on their activities in applications with a large number of users using in-memory processing. We present a new instantaneous behavioral model to examine users' activities and actions rather than results of their activities in order to analyze and predict their behaviors. For the purpose of this research, we designed a software to simulate user activity data such as users' swipes and taps, and studied the performance and scalability of this architecture for a large number of the users.