Geo-coded aggregate sentiment analysis of Tweets

When working at the Deloitte, I collaborated with one the world's largest fast-food chains to perform sentiment analysis on their social media data. This resturant chain has significant locations across North America and needed a way to visualize sentiment in various marketing regions. Using Tweets from GNIP (a historical repository of Tweets), we were able to enrich each with a sentiment score of -1 to 1 with machine learning. From there, we rendered aggregate mood scores across regions (fig. 1), visualizing scores with negative and positive heatmaps. During the design phase, we worked with the client to tease out current marketing frustrations. This process led to opportunities for proactive advertising and employee education opportunities.