Detroit Soup is a local, community based grant funding organization with a very interesting funding model (read more here). Local organizations get up in front of a community gathering, pitch their idea, and then over a soup dinner, the audience discusses the proposals, and votes on their favorite. Admission fees (and any matching $$) is then given to the winner towards making their idea happen.
Working with the always awesome Amy Kaherl from Detroit Soup, and my great Detroit School of Data Science colleague, Kat Hartman, we collated and organized data from grant proposal submissions for Soup. Our goal was to help Amy understand a bit more about about trends in Soup events, who is proposing ideas, and what kind of ideas are coming forth. In addition to less exciting bar graphs and trend lines, I took the text input discussing each proposal’s project summary and plans, and used Python to cluster the proposals. Although this work is relatively simple (bag of words approach with nltk, kmeans clustering) I will be working with Amy to move beyond this naïve method and pick several areas of interest to find the relevant proposals.