LINEAR REGRESSION MODEL IN CONTEXT OF MOBILE APPLICATIONS USAGE

Abstract

The popularity of mobile applications has resulted in an ever-increasing number of programmes being installed on smartphones. Whether or whether it is possible to predict which app a user will open is the subject of this study. The ability to forecast what apps will be needed in the future can aid in pre-loading the required apps into memory or in floating the relevant apps to the home screen to speed up launch times. We analysed a wide range of contextual information from the MDC dataset, including the user's profile, time and location, and the most recent App they utilised. The findings of our investigation can be divided into three categories. First and foremost, contextual information may be utilised to better understand how a user interacts with an app and to make more accurate predictions about how that app will be used in the future. A large part of the forecasting accuracy for the MDC dataset comes from the correlation between the sequentially used applications. The linear model is better than the Bayesian model because it can take into account all of the relevant information and provide more precise predictions than the latter. Predictions about app usage based on contextual information such as time, location and user profile and the most recently used app have been offered as a consequence of our research. For app usage prediction, we studied the topic of context awareness, and one of the things we observed was that context can significantly affect a user's app usage behaviour. We can deduce some patterns about how mobile app users interact with the software by examining contextual data. Personal mobile systems that employ contextual information to dynamically offer information, such as Apps to be used, to the user and improve the user-mobile phone interaction experience are suggested by this work. A personal mobile device is an example of this type of system.