Unsupervised Exploration of craigslist musicians communities in the United States.

• Click here to see the Flask app I built.
• Click here to see the blog post I wrote explaining my workflow for this project.

I scraped craigslist musicians sites with over 300 posts to get the text of each post. After preprocessing and text vectorization using tf-idf, I used non-negative matrix factorization (NMF) to model topics for the posts. The following files and folders are organized so that:

• pres.pdf contains my findings from data exploration and NMF
• /notebooks contains jupyter notebooks which demonstrate my workflow from preprocessing > modeling > visualization
• app.py is a Flask app I created to visualize the scraped posts according to their topics using an interactive leaflet map and dc.js libraries for charting. After cloning, run python app.py from the root to use the app which should then be running on 0.0.0.0:5000.
  The app looks like this: