[kevinpet]

This is essentially a "job market paper" for a data science bootcamp for recent PhDs.

The contents are nothing groundbreaking but just an exploration of how different models and data sources can be used to predict something. Insofar as the purpose is to demonstrate the author could do the same sort of analysis about questions like will customers buy your product based on unstructured reviews posted on amazon, I think its successful.

The only problematic aspect seems to be the notion that you can "recommend" a form of birth control based on modeling preferences. This is definitely a Jurassic Park type of article, interesting because it's so clear that data science built on self reported information on reddit is a bad way to actually recommend options when there are much better method available (e.g. professionals).