Title: Succeeding in Mobile Application Markets
Presented By: Dr. Ehsan Noei
Abstract: Applied machine learning and knowledge discovery have become essential parts of the software development industry in the past few years. In this talk, we target the application of data science and machine learning for developing and maintaining a successful Android app on Google Play Store, where receiving high star-ratings and achieving higher ranks are two primary elements of success. First, we discuss the importance of mobile device metrics, such as screen size, as opposed to app metrics, such as user-interface complexity. With the knowledge of critical device metrics, developers would be able to prioritize their testing activities with respect to certain devices. Second, we identify the topics of user-reviews (i.e., users' feedback) that are statistically significantly related to star-ratings using proportional marginal variance decomposition approach. Therefore, developers can narrow down their efforts to the topics that are statistically significantly related to star-ratings, rather than statistically trivial, although frequent, topics. Third, we propose a solution, with a precision of 79%, for mapping user-reviews (from Google Play Store) to issue reports (from GitHub). Fourth, by applying our mapping method and learning from successful apps, we provide developers with a solution for prioritizing user-related issue reports. Finally, we determine the rank trends in the market over time. We also investigate the most statistically important factors that are related to changes in the ranks, such as release latency, using a mixed-effects model. The outcome of this research helps app developers to become more successful in the competitive market of mobile apps. It also shows the vital role of data science and machine learning in modern software development paradigms.
Biography: Ehsan earned his PhD from Queen's University working with Prof. Jenny Zou. For his PhD thesis, he applied various data analysis and machine learning techniques in mobile application development domain. His research has been published in top-tier software engineering journals and venues, such as IEEE Transactions on Software Engineering (TSE), Empirical Software Engineering (EMSE), and Foundations of Software Engineering (FSE). Ehsan's primary research interests include data science, machine learning, and automated software engineering and re-engineering. He is currently a Postdoctoral Fellow at the University of Toronto working with Prof. Kelly Lyons and Prof. Periklis Andritsos on a data-driven knowledge mobilization project.