Predicting Mobile Application Usage Patterns Using Linear Regression: A Data-Driven Approach

Abstract

One central aspect on the usage patterns of mobile applications was on the prediction aspect that has become important since mobile applications have become central to individuals in the current world with digital caricatures. This is the correct forecast of how people are going to use mobile applications and it will assist the App developers and marketers a long way to improve their user touch and programming policies. In the current paper, the linear regression approach to forecasting the usage trend of mobile applications is utilized based on the data-driven technique. The existing study builds on the study by Smith et al. (2018) who examined the user behavior (in the mobile environment) and certain user-based product aspects, such as the frequency/period/and demographics of the user to which a predictive model will be applied. Using an actual interactions data with mobile application, the performance of the model is quantified in some of common measures of regression, which include mean squared error (MSE) and R-squared (R 2 ). The results confirm that GRs are adequate to capture meaningful usage patterns but there could be certain usage patterns, which may require more intelligent regression models. The article identifies the strengths and shortcomings of the linear regression to forecast the mobile utilization and sets the implications of any further research on the same topic. The recommendation regarding future is that advanced machine learning algorithms would generate improved results of prediction and could be connected to more contextual data sources.