Photo by NeONBRAND on Unsplash
One of the most common questions I get is:
Should I attend a data science Bootcamp or Masters?
I usually recommend people avoid these specialized data science programs because they tend to lack the rigor to prepare students for the quickly evolving world of data science.
I think these programs are extremely expensive in an attempt to capitalize upon the data science craze. And, thus, are more focused on the money than preparing students.
I wasn’t sure if I was overly biased, though, so I posed the question on LinkedIn. I was blown away by the number of responses. They were so good, in fact, that I wanted to try and provide a summary and some additional thoughts.
Overall, though, most people seemed to agree that these programs usually disappoint and don’t provide the necessary elements to be a successful data scientist.
A Solid Foundation is Key
By far the largest trend I saw emerge was the importance of building a foundation based on math, statistics, and computer science.
Tyler Byers, a lead data scientist at RiskLens, said the following:
I got halfway through a Master’s in DS. I was *very* disappointed with the lack of rigor compared to various MOOC options out there, especially considering the price point (edX, Coursera, Udacity have been my go-tos).
It seems that these specialized programs often don’t focus on building a strong foundation of skills, but rather try and teach the hottest areas with just enough knowledge. For example, being able to piece together enough Keras code for image classification.
While I think everyone needs to start somewhere, if your learning stops there, you will be underprepared to enter the field of data science.
As Joe Reis said:
A solid math background helps a ton. I’d say learn the fundamentals — math, stats, CS.
I couldn’t agree more. It feels like the state-of-the-art is changing on a weekly basis and if you don’t understand fundamentally how these models work, you won’t be able to adapt and adjust as things change.
It is more akin to giving someone a fish instead of teaching him or her to fish. You might leave a specialized program with some additional skills (the fish), but those skills quickly show a lack of depth when entering the workforce.
That being said, if you find a specialized data science program that offers strong foundational courses such as statistics, calculus, and linear algebra, it could be excellent.
In fact, Joe Hoeller found value in Udacity’s nano-degrees:
Udacity has a nano degree program that focuses on math just as much as it does w framework APIs. I enjoyed it and learned a lot.
On the other hand, if you all you see are courses on high-level subjects, you might look elsewhere.
Portfolio of Work
One area in which these types of programs seem to do well is making sure each student has a portfolio of projects at the end. This is great, but you might be able to achieve the same result for a lot less. As Matt Harrison said:
You need something that will give you a portfolio to talk about. If you are highly motivated you can do that for free. 💰⌛ Most people are not highly motivated and thus need to pay money to feel motivated. Both an MS and a bootcamp can give you a portfolio. (The reverse is also true). 🤔🏋️🐍
I completely agree. There are a ton of amazing courses online which will help you gain the skills you need to build a great portfolio. And most of them are free! The one thing you don’t get from self-learning, though, is an association with a respected program. While it seems like that shouldn’t matter, it can be critical to get past an initial resume screen.
I think this is another area in which specialized programs fail. Most are not associated with a well-known university and can more easily be overlooked by recruiters even if the program is good. Interestingly enough, the few that are associated with prestigious universities seem to lack significant rigor unless they are part of a more traditional degree such as statistics or computer science.
Also, while you might gain a portfolio of work from a data science bootcamp, I have found that the understanding of the project is often superficial. Usually, they require you to do enough to get it working, but not to actually understand the inner workings of what you developed.
Soft Skills Matter
Ryan Russon made a great point on the importance of soft skills:
I would say that some of the most valuable experience I had was in the US Navy, because that is where I truly learned to work effectively with people and their expectations, which is where I believe much of the true value of data science lies.
Almost all technical programs miss this important point and offer very little to help students develop soft skills. I think, though, that this could be an area in which specialized programs could have an advantage.
They seem to be trying to be more well-rounded and I have seen some with courses more focused on leadership and communication. This is great! But usually comes with a lack of foundational technical knowledge. If only there were a program that provided solid foundational knowledge as well as soft skill training.
Lastly, it was clear from the comments and thoughts that regardless of the program you choose you have to stay curious and always be learning.
Learning doesn’t end with your formal education.
So, no matter how you choose to learn about data science, I hope you can role your passion for the area into a life of learning.
In general, though, if you are looking to pursue more education, I would make sure whatever program you choose offers rigorous foundational knowledge, a portfolio of work for which you have deep knowledge, and as a bonus — helps you build your soft skills.
Unfortunately, it seems that most specialized programs are currently missing the mark.
You can see all of the comments on the subject here.
Join my email list to stay in touch.