The world’s leading publication for data science, AI, and ML professionals.

Should I Attend A Data Science Bootcamp?

Three Questions To Help You Decide Whether To Take The Leap

Photo by Glenn Carstens-Peters on Unsplash
Photo by Glenn Carstens-Peters on Unsplash

Going to bootcamp is a big decision. The higher quality ones are generally full time and require a significant financial investment (both in terms of direct cost and foregone wages).

So here are 3 questions that you should have clear answers to before you take the leap:


1. Do I have the financial reserves and the discipline to handle 7 months of unemployment?

Besides the bootcamp, which typically lasts 3 months, you should also plan for 3 to 4 months of job hunting afterwards. The lucky few will have jobs lined up by the end of the bootcamp or immediately afterwards. But they are the exceptions not the rule – more typically, you will spend at least a few months post bootcamp hunting for that elusive job.

And during that time, you will no longer have the daily structure and schedule of the bootcamp. If you are not a self-starter (and a natural optimist), this period of time can be extremely draining due to the uncertainty and rejection. Moreover, if you didn’t prepare a sufficiently large financial cushion, then that’s a second layer of pressure during an already stressful time.

I’m not saying that you should not go. I thoroughly enjoyed my experience as part of the METIS bootcamp, and I love the job that I got following the bootcamp. But changing Careers is no joke, and you should expect to work hard and experience some adversity.


2. Am I OK with a Data Analyst title?

Let me preface this by first saying that data analysts are great. They are well paid, work on interesting things, and contribute a lot of value to the businesses that employ them. But one common thing I heard from other Metis bootcamp attendees was, "I didn’t go to a Data Science bootcamp to become a data analyst."

But the truth is that a large portion of bootcamp grads become data analysts. There just aren’t enough data scientist jobs out there to meet the demand of the ever growing supply of newly minted data grads. And on top of that, some companies are realizing that they may be better served by hiring 1.5 (or even 2) data analysts rather than a single more expensive data scientist.

There is also a large skills and responsibilities overlap between data scientists and analysts. What Company A calls a data scientist, Company B might call a data (or product) analyst. The overlaps are on things you might expect like mastery of SQL, data visualization skills, the ability to clearly communicate quantitative insights to a non-technical audience, and a baseline knowledge of modeling and statistics in Python (or R). Where the roles might differ are:

  • The amount and depth of Python/R knowledge required. And the amount of time you spend coding away versus meeting with your stakeholders.
  • Whether you work more on slicing data for insights or attempting to model and predict uncertain outcomes.
  • Whether your work is more closely aligned with week to week business needs or more focused on longer term projects and research.

I would recommend not fixating too much on the title. Rather, if the company and the role sound interesting, then go for it. Granted, if you are dead set on applying state of the art deep learning techniques to the mysteries of the universe, then you should hold out for a data scientist (or machine learning engineer) role. But if that’s the case, you would probably be better served going to graduate school versus a bootcamp.


3. Do You Respect The Model Building Process?

Data science is not like other software development jobs. You can spend a lot of time putting together a dataset, cleaning it, and building a model only to find that it predicts badly or explains very little.

There is no guarantee that your time spent building will result in something worthwhile. And when it doesn’t, there will be temptation to game the system by either overfitting or training the model on your test set (or both). As an aspiring data person, you need to be able to resist these temptations.

Overfitting might make your model look better in the eyes of your peers, but it also severely hampers your model’s ability to perform well on real out of sample data. This question applies not just towards whether to go to bootcamp or not, but also whether to pursue a career in data science period.

Do so only if you are OK with unsteady and nonlinear progress in your work and are willing to stay true and honest to the modeling process – which means running as honest a backtest as possible for your predictive models, not making modeling decisions using your test data, and reporting your results regardless of whether they were world beating or not.


More Data Science Related Posts By Me:

What Do Data Scientists Do?

How Much Do Data Scientists Make?

Got Data Science Jobs?

A Better Way to Skill Up

Making Data Science Interviews Better


Related Articles