Personal and social predictors of use and non-use of fitness/diet app: Application of Random Forest algorithm

Abstract

This study investigated the various groups of factors that predict individuals’ use and non-use of fitness and diet apps on smartphones. Unlike previous research on fitness and diet apps which have mainly studied individuals’ intentions to use the apps, this study focused on the prediction accuracy of various factors that lead people to use fitness and diet apps through analysis of data collected from users as well as non-users of these apps. To examine prediction accuracy, this study applied the Random Forest algorithm. According to the findings, prediction accuracy higher than that of 70 percent was observed for nine factors: age, annual income, education, perceived obesity, dieting efforts, number of smartphone apps currently used, daily time spent with smartphone apps, perceived benefits from exercise, and social influence. A major contribution of this study is its detection of those factors predicting actual behavioral decisions regarding use of fitness and diet apps, as opposed to future intentions..