Predicting Mobile Apps Performance using Machine Learning
Journal of System and Management Sciences · 2022
The number of mobile applications is expanding daily along with new upgrades.It might be difficult for developers to identify the key factors that affect an application's rating.This paper aims to help application developers in improving their decisions and adding new features to mobile applications.2.24 million Records from the Google Play Store were used in the research to create a dataset for the purpose of predicting the performance of mobile applications.The dataset included different features such as rating count, minimum installs, category, maximum installs, and price; add support, in app purchasing, total size and content rating.Four machine learning methods were used: decision trees, naive bayes, ANN, and random forest.The random forest outperformed the other algorithms in terms of precision, recall, and f-measure.Additionally, the top three features influencing the performance of the app were category, minimum installations, and rating count.This research offers practical recommendations for tech entrepreneurs and app developers to better comprehend the features affecting the performance of the mobile apps and give them the ability to predict the app performance based on their assumptions of the number of installs, price, category, and other features, which will help them properly, plan their development initiatives to avoid failure.