Analysis for music royalty portfolio origination.
I undertook this work one weekend in 2021 as part of an application to work at a fund primarily focused on the buying and selling of music royalties.
As the fund no longer exists, I have uploaded my work as a portfolio piece. Below, I provide commentary on various pieces of the work for those interested.
In short, the ask of the assessment was to investigate the "one hit wonder" hypothesis.
Within that, the exercise covered various activities in the data science skillset:
- Procuring and processing data (web scraping, API calls, and the like)
- Analysis (exploration, basic modelling)
- Write-up (making sense of the data)
The repository is laid out as follows:
- scripts: scraping and ingestion scripts (essentially, the "ETL" bit)
- notebooks: the analysis
- outputs: the final output report
- Standard Python Data Science stack (pandas, sklearn, matplotlib)
- SQL (sqlite)
- Scraping / API (httpx, beautifulsoup)
- If you have any feedback about this, I'd be glad to hear it. Send a message across.
- Looking back, I think the analysis is somewhat overcooked - we could reduce the number of one hit wonders by being more conservative (e.g. limiting to top 10).