Both of us attended the Brown Datathon at Brown University a few weeks prior, and this sparked our interest in the field of Data Science. We didn't have too much success at that event, so we decided to give it another shot at HackRU.
What it does
We used various charts to visualize a Google Play Store App data set. This data set contained information such as reviews, rating, category, and even the polarity and sentiment of a user's review for many different apps available on the play store. We also created a model to predict an app's future rating given its attributes and a sample of the sentiment of its reviews.
How we built it
We did our preliminary exploration of the data sets using jupyter notebook, cleaning and transforming the data, isolating important and relevant features that would be used in our visualization and rating prediction model. On the frontend, we used crossfilter for dimensional data analysis and d3.js and dc.js for an interactive visualization of the results.
Challenges we ran into
The biggest problem that we ran into was creating the prediction model for the data set. Not only does the complexity of creating linear regression increase exponentially with the number of features, but we also had trouble finding significant correlations between many of the data points.
Accomplishments that we're proud of
We had little experience in the field of data science and creating prediction models, so we're proud of ourselves for creating a working prediction model. We are also proud of the interactive visualizations we created, as they really make you think about relationships in the data you might not have thought about before.
What we learned
As a group passionate and interested in the field of data science, we are glad Gartner gave us another opportunity to work with data once again and garner more experience in this field. We learned about how to clean and manipulate data properly, and the experience with creating visualizations will surely help us in our future data science related endeavours.