Since the launch of modern smartphone in 2007, the mobile phone industry has proliferated. Over 3 billion people spend substantial amounts of their day using apps on their smartphones with global internet penetration standing at 57%. Today, there are almost 2.46 million apps available for download in Google Playstore, 1.96 million in Apple App Store, 700,000 in Windows Store, and 479,000 in Amazon Appstore. Although many have seen mammoth success, most of them turn out to be unsuccessful. Of paid apps, about 90 percent are downloaded less than 500 times per day — and earn less than $1,250 a day. Moreover, 80% of them received less than 100 downloads. While competition in the app market is high, failure isn’t always a result of getting lost in the noise. In most cases, there are other contributing factors. Poorly researched market and audience would most likely top that list.
In this project, we will explore the Worldwide Mobile App User Behavior Dataset to predict if a users pays for apps or not. We will also use novel model interpretability libraries like SHAP, Eli5 and Permutaion Importance to further understand the predictions and the features responsible.