Shreyam Kela, Harsh Agarwal, Vaishali Koul, Sayalee Shankar Bhusari
Description: Ever made an app that got you pulling all-nighters for weeks, only to get the lowest possible ranking on the app store? Feels terrible, right? Next time you build an app, use this prediction system to assess the success of your app before you start banging away at your keyboard to get that app on the app store. Our prediction system analyses the app description you provide and it provides you back with the detailed analysis of your app's potential selling quotient.
Methodology: Apply Natural Language Processing on the App Store dataset to find the K-Nearest Neighbours of the new description entered, using TFIDF of the descriptions in the dataset and their Cosine Similarities with respect to the new description. Our system analyses the new description entered and determines the probable outcome as success or failure and the detailed analysis such as the potential app store rating and so on.
Dataset: https://www.kaggle.com/ramamet4/app-store-apple-data-set-10k-apps
Professor's Comment: I would also predict key metrics such as number of downloads by geography and age group etc..
Design Mockup: https://invis.io/C7P3GP56X5K