Predicting Users' Demographic Features Based on Searched Queries and Installed Apps and Games

Abstract

Employing various strategies to catch interest in using online app stores has become a common trend in recent decades. One of the dominant factors in determining the success of online businesses in this area is whether they have information about users' demographic features such as gender or age. In this study, we try to detect these features based on the lists of installed applications, installed games, and searched queries collected from an Iranian mobile application store. For this goal, we use a wide range of machine learning techniques to identify which model has the highest performance in these classification and regression tasks. Our findings show that we can detect genders with a balanced accuracy of 0.76. We also achieve 10.02 as the RMSE for age predictions and the ROC AUC of 0.81 in determining users' age groups.