<<<<<<< HEAD

1. I worked on the Mobile Apps Usage dataset from iTunes Search API at the Apple Inc website. The dataset consisted of data of 7200 Apps across 23 genres. My goal was to look at the most popular Genres, most popular Apps within those genres and whether a user will pay for a particular App.
2. I have used Linear Regression model to predict the features that most influence the user rating for each App genre.
3. I have also used DBSCAN and HDBSCAN Clustering algorithms to understand the grouping of data across the genres, but did not reach a conclusive result.
4. Gridsearch was used with ElasticNet, RidgeCV and LassoCV and ElasticNet gave the best score with the conculsion that number of ratings for current version is a good predictor of user ratings. But this inference could not be confirmed by any other algorithm.
5. Pipelines were used 'rbf' kernel gave the best score. Again, the conclusion needs further investigation.
6. For Text Analysis, only English language stop words were used. The modeling techniques used were KNN and Linear Regression. ======= I worked on the Mobile Apps Usage dataset from iTunes Search API at the Apple Inc website. The dataset consisted of data of 7200 Apps across 23 genres. My goal was to look at the most popular Genres, most popular Apps within those genres and whether a user will pay for a particular App.

167edfbd85aa75460dd4e335d6daf14797f6643d