Pretty sick of my postdoc and would like to find stable employment in the private sector. Thinking about data science.
Anyone here made this transition or know anyone who did? What was their skillset like? Did they have to take additional data science courses before being considered hireable in those kinds of roles? Basically, how well does being a quant political scientist prepare you for transitioning to data science?
Data science is a broad field. Shortest path is getting your foot in the door, some entry level job. Where you can start working with SQL, Python, at least some machine learning, etc. It takes about 7 years to pick up the skills that will land you the 200k a year job. So it's not easy, but it is lucrative.
Learn Python. I was surprised by how little attention R gets within Data Science. That being said, the stats skills of most data scientists is extremely low, since they essentially don't do hypothesis testing.
You're right that industry based data science is more about prediction than hypothesis testing, but that definitely does not mean most data scientists have low stats skills. They just have a different skill set than what we learn in poli sci since they mostly care about accurate predictions rather than accurate inferences about how one variable relates to another.
Like others have said, learn python and at least some sql. One of the bigger differences in my experience is the use of non-linear algorithms that we don't use a lot in poli sci (at least they weren't discussed in my grad program) like decision trees/random forests, KNN, and neural networks, as well unsupervised stuff like clustering that I also don't remember seeing a lot of in poli sci except for like text analysis stuff. Datacamp provides a decent starting point but obviously won't take you as far as actual textbooks, but there are tons of free pdf textbook you can find through a Google search.
Also be sure to know the difference between data analyst, data scientist, and data engineer. With a phd in poli sci all you need to be qualified for data analyst positions is to learn the preferred language of whatever company you're applying to, but you will mostly be calculating group means and other descriptive stuff. For data science positions, you'll need to know the algorithms mentioned above. For data engineer positions, you'll need to learn a lot of other programming stuff, things Amazon Web Service stuff and Apache Spark.
Starts w/ assertion that is questionable at best: "Most professors have never been gainfully employed"
How does he know that? Anecdotes? Systematic data? Just him? If we want anecdotes, a vast majority of my cohort in grad school had been "gainfully employed" before starting their PhD (and I'm not talking about bagging groceries -- not that there's anything wrong with that). I'm looking at my faculty, and from what I know, I'd say a majority have held a job at some point, either before grad school, during grad school, or after grad school.
Anywho. Not a huge thing. I'm sure he's a better data scientist than opinion writer.
I did. It was fairly easy because as political scientists we actually work with data and know where it comes from, unlike stats or CS phds. I had to learn python. I actually do causal inference (not ML prediction) so its pretty similar to my PS work.
My total comp is around 150k.