A gender-specific behavioral analysis of mobile device usage data

Mobile devices provide a continuous data stream of contextual behavioral information which can be leveraged for a variety of user services, such as in personalizing ads and customizing home screens. In this paper, we aim to better understand gender-related behavioral patterns found in application, Bluetooth, and Wi-Fi usage. Using a dataset which consists of approximately 19 months of data collected from 189 subjects, gender classification is performed using 1,000 features related to the frequency of events yielding up to 91.8% accuracy. Then, we present a behavioral analysis of application traffic using techniques commonly used for web browsing activity as an alternative data exploration approach. Finally, we conclude with a discussion on impersonation attacks, where we aim to determine if one gender is less vulnerable to unauthorized access on their mobile device.