With three materials science students and a computer science students in the team, we were set on taking on a challenge involving the interpretation of mass amounts of data using machine learning, as well as the communication of said data.

What it does

Hypothesis: Customer Sentiment in Tweets is extremely valuable and can be used as a guide for the health of a company in terms of economic growth from stock analysis. We have also included a built in twitter bot to listen for tweets containing certain key phrases such as "jetblue", evaluate the customer sentiments as well as provide GUI functionality to directly retweet or reply to high sentiment tweets.

How we built it

We started off with analyzing JetBlue consumer data from Twitter in correrlating consumer sentiment to economics. We discovered a clear linear relation between the 2, and continued to explore our hypothesis with more data from other databases as well as from other competing airlines.

Challenges we ran into

Aside from the usual challenges of sitting for 2 days, we encountered most of our challenge in data acquisition and filtering for interpretation. Most open source domains offer little or no useful consumer data, and the databases that do have a steep monetary cost. We ended up spending majority of the time acquiring data from Twitter and Reddit and filtering for usable consumer feedback, and Stock databases to compare and correlate our findings from social media and economic trends. Once the data was acquired and was filtered to our satisfaction, we were able to use Google Cloud Sentiments API to quantize consumer feedback and create data visualization models in python and R.

Accomplishments that we're proud of

We are proud in our process of data scraping/processing so we can confidently provide evidence to support our hypothesis. The data found was processed to provide intuitive visualizations that clearly show a strong correlation between economic trends and consumer sentiments. We are happy to provide an integrated GUI application with in-situ data analysis methods as well as direct functionality to our resident HEAVENS DOOR consumer relations twitter bot.

What we learned

Big data analysis is a slippery slope to hell. There's a dangerous wealth of information available to virtually anyone with an interest and an internet connection.

What's next for Heavens Door


Built With

Share this project: