Modeling User Engagement in Mobile Applications Using Machine Learning

Abstract

The application of machine learning for modeling user engagement in digital applications is rapidly emerging as a key area of research. A dataset of behaviours of mobile app users was prepared, which included such parameters as time spent in using the app, battery drain rate, age and gender, as well as the number of installed apps.  Predictive models were created using Random Forest and Gradient Boosting as ML approaches to suggest tactics that engaged and enhanced the user further. The findings were that the best accuracy estimation was 92%, attained by the Random Forest model and hence contributed to a rich understanding to improve different aspects of app usage and user experience