Predicting Mobile Game Success Using Data Analytics

Loading...
Thumbnail Image
Date
2017-11
Journal Title
Journal ISSN
Volume Title
Publisher
The British University in Dubai (BUiD)
Abstract
Since the advent of arcade games and the development of the Wireless Application Protocol (WAP) at the close of the millennium, the mobile game app industry has exploded; and subsequently has transformed the ideologies of mobile technology and software developers to forward thinking within the dimension of innovative mobile game development. After the first decade of the new millennium has passed, and even though billions of dollars in revenue have been realized from mobile game apps, there is still a gap in literature with regard to mobile game user behavior and methodologies for predicting the likely success of mobile game apps during the development phase. Game features and ARM strategies are analyzed and discussed as primary drivers of mobile game app success. This study addresses these challenges through data driven research of the mobile gaming application market, mobile gaming application features, user acquisition and retention trends, and monetization strategies using the CRISP-DM model for data mining in order to prove a successful method for predictions of mobile game application success. The attainment of the prediction of one mobile game app from a sample of 50 was accomplished by running a batch prediction for the game features dataset, and a separate batch prediction for the user behavior dataset. The lists were then integrated, a final list of games which appeared in both lists was generated for further comparison. According to the prediction model results for the dual datasets, the most successful mobile game app from the 50 game sample was Game of War-Fire Age; the most successful genre was Puzzles, and the most successful developer was EA Sports. Where success is described based on the best match with the results of the study. The most successful game predictions were extracted and compared to the predominating user behaviors for further analysis and implications. Significant outcomes for the comparisons included the predominance of the Social Networking features, Offers, and IAP 90% to 100% of the sample. A model of mobile game app success prediction based upon the game features values that are created proposed.
Description
Keywords
Mobile games., monetization, location-based mobile apps, predictive analysis, CRISP-DM
Citation