I gave this presentation on December 6, 2021 as part of the course SURV727 “Fundamentals of Cmputing and Data Display.” I use R for data collection, and then clean and analyze the data in Stata.

My data science experience grew quickly and greatly during this course, and I had a lot of fun combing various sources of data involving API’s from Spotify, Last.FM, Wikipedia, and Google, and using web scraping techniques to obtain review scores from Wikipedia.

A technique I used was to funnel strings through search engines in order to obtain data from the API. I almost think of it as a “fuzzy match” where we let the search engine handle the fuzziness instead of creating own string comparators.

The results are very rough, but I think the process for collecting the data is very interesting, and there are plenty of code examples. If you take the results on face value (you probably shouldn’t), the model predicts an album with 1.7 million listens could earn about 100,000 additional listens by improving their album from a 7/10 to an 8/10. This translates to about $300 USD if you assume revenue of about $0.003 USD per listen.

• Data Science
• R
• Stata
• Coursework
• Research