Dan Vatterott

Data Scientist

Psychology to Data Science: Part 1

A number of people have asked about moving from a PhD in Psychology/Cognitive Psychology/Cognitive Neuroscience to data science. This blog post is part of a 2-part series where I record my answers to the best and most common questions I’ve heard. Part 2 can be found here.

Before I get started, I want to thank Rick Wolf for providing comments on an earlier version of this post.

This first post is a series of general questions I’ve received. The second post will focus on technical skills required to get a job in data science.

Each header in this post represents a question. Below the header/question I record my response.

Anyone starting this process should know they are starting a marathon. Not a sprint. Making the leap from academia to data science is more than possible, but it takes time and dedication.

Do you think that being a Psychology PhD is a disadvantage?

I think it can be a disadvantage in the job application process. Most people don’t understand how quantitative Psychology is, so psychology grads have to overcome these stereotypes. This doesn’t mean having a Psychology PhD is a disadvantage when it comes to BEING a data scientist. Having a Psychology PhD can be a huge advantage because Psychology PhDs have experience measuring behavior which is 90% of data science. Every company wants to know what their customers are doing and how to change their customers’ behavior. This is literally what Psychology PhDs do, so Psychology PhDs might have the most pertinent experience of any science PhD.

When it is the right time to apply for a boot camp?

(I did the Insight Data Science bootcamp)
Apply when you’re good enough to get a phone screen but not good enough to get a job. Don’t count on a boot camp to give you all the skills. Instead, think of boot camps as polishing your skills.

Here is the game plan I would use:
Send out 3-4 job applications and see if you get any hits. If not, think about how you can improve your resume (see post #2), and go about those improvements. After a few iterations of this, you will start getting invitations to do phone screens. At this stage, a boot camp will be useful.
The boot camps are of varying quality. Ask around to get an idea for which boot camps are better or worse. Also, look into how each boot camp gets paid. If you pay tuition, the boot camp will care less about whether you get a job. If the boot camp gets paid through recruiting fees or collecting tuition from your paychecks, it is more invested in your job.

Should I start a blog?

Yes, I consider this a must (and so do others). It’s a good opportunity to practice data science, and, more importantly, it’s a good opportunity to show off your skills.

Most people (including myself) host their page on github and generate the html with a static site generator. I use octopress, which works great. Most people seem to use pelican. I would recommend pelican because it’s built in Python. I haven’t used it, but a quick google search led me to this tutorial on building a github site with pelican.

I wish I’d sent more of my posts to friends/colleagues. Peer review is always good for a variety of reasons. I’d be more than happy to review posts for anyone reading this blog.

How should I frame what I’ve done in academia on my CV/resume?

First, no one in industry cares about publications. People might notice if the journal is Science/Nature but most will not. Spend a few hours thinking about how to describe your academic accomplishments as technical skills. For example, as a Postdoc, I was on a Neurophysiology project that required writing code to collect, ingest, and transform electrophysiology data. In academia, none of this code mattered. In industry, it’s the only thing that matters. What I built was a data-pipeline, and this is a product many companies desire.

We all have examples like this, but they’re not obvious because academics don’t know what companies want. Think of your data-pipelines, your interactive experiments, your scripted analytics.

Transforming academic work into skills that companies desire will take a bit of creativity (I am happy to help with this), but remember that your goal here is to express how the technical skills you used in academia will apply to what you will do as a data scientist.

Many people (including myself) love to say they can learn fast. While this is an important skill it’s hard to measure and it calls attention to what you do not know. In general, avoid it.

Did you focus on one specific industry?

I think a better question than what industry is what size of team/company you want to work on. At a big company you will have a more specific job with more specific requirements (and probably more depth of knowledge). At a smaller company, you will be expected to have a broader skill set. This matters in terms of what you want in a job and what skills you have. Having industry specific knowledge is awesome, but most academics have never worked in an industry so by definition they don’t have industry specific knowledge. Unfortunately, we just have to punt on this aspect of the job application.

Anything to be wary of?

No matter what your job is, having a good boss is important. If you get a funny feeling about a potential boss in the interview process, don’t take the job.

Some companies are trying to hire data scientists but don’t want to change their company. By this I mean they want their data scientists to work in excel. Excel is a great tool, but it’s not a tool I would want to use every day. If you feel the same way, then keep an eye out for this.