Guide to Data Science Bootcamps in 2019

In our new, data-driven landscape, it’s not uncommon for people who work in tech or looking to switch to a different industry to reskill on their own and learn how to work more productively. Data Science bootcamps are sometimes intended to fill knowledge gaps, but many programs teach basic fundamentals and build up to advanced topics. As a data scientist, you’ll be using a variety of programming languages, such as Python or R, and visualization tools like Tableau to help organizations make data-driven decisions.

Looking for a quick way to compare programs? Check out our list of data science bootcamps. We cover everything you need to know – from location and education requirements, to job placement figures and curriculum details.

Learn More About Bootcamps

What are Data Science Bootcamps?

Bootcamps are intense educational programs that pack critical data science skills and technologies into a short period of time. Established bootcamps, like Flatiron School and General Assembly, offer in-person, online, or a hybrid approach that combines online and campus learning, whereas Thinkful provides a completely tailored online learning experience. Many schools will work with your schedule or offer part-time programs. Part-time data science bootcamps may take more time to complete, but in some cases, courses are offered in the evening, which can be very appealing to students with full-time jobs.

Some data science bootcamps (e.g. Metis, NYC Data Science Academy, and Galvanize are aimed at those with experience in object-oriented programming language, statistics, databases, or math. The free Insight Health Data Science: Fellows Program is offered to PhD or MD graduates transitioning to big data jobs at leading healthcare organizations. Other bootcamps offer programs for all skill levels. For example, Microsoft’s DS3 summer school hopes to convert curious college students into data scientists.

Are Data Science Bootcamps Worth it?

Before you commit to a specific program, consider the bootcamp’s instructor and mentor quality, cohort makeup, curriculum, portfolio work, and outcomes rates and support. Data science is skills-based, so you should be looking for instructors who have serious, hands-on experience in your preferred fields. You may want to look for bootcamps that focus on group learning as well and admit a full cohort of students with diverse backgrounds.

It’s also important to decide how much structure you want. Are you comfortable with self-paced learning? Or do you prefer deadlines, assignments, and homework? If you’re new to data science, check out beginner and intermediate levels, which often include lectures and projects in fundamentals (e.g. Python). If you’re looking at advanced fellowships, you’ll have more independence to focus on what you’d like to work on and build your portfolio.

For data science jobs, your portfolio is your real resume – employers want to know what you’ve done, why you did it and how it’s original. For example, Thinkful helps its Python learners develop an active GitHub portfolio; NYC Data Science Academy asks students to spend the last two weeks of camp on a capstone project. With the ASI Data Fellowship, you have the chance to work with industry partners on real-world problems. A strong bootcamp will help you build a diverse portfolio that’s also geared towards what employers look for in their candidates.

Many bootcamps have set themselves up as talent pipelines, funneling trained data scientists to eager partner companies. If you’re hoping to land a job after you graduate, look for programs that provide plenty of career training. For example, Metis sets up mock interviews, company site visits and consulting projects. General Assembly provides an in-house career coach and instruction in salary negotiations. You may also want to look for a bootcamp that provides alumni with post-graduation support, which sometimes includes alumni networking or emails with relevant jobs for recent graduates.

Most bootcamps advertise job placement rates (some boast rates of 100%), but these percentages are only part of the story. You may only receive job offers from the bootcamp’s partner companies, not end up in a “true” data scientist role, or wait three months before receiving an offer. It’s also possible the starting salary is a lower median salary than you expected. Take a careful look at what job placement entails and make sure the career they’re preparing you for is the career you want.

Choosing the Right Bootcamp

1. Establish Your Career Goals

Where do you want to be in five years? Do want to get a data science job straight out of bootcamp or do you just want a skill set that will allow you to work on independent projects? Knowing your goals will help you narrow your options. If you’re lost in a dead-end data role and not sure what your next step is, you might want to explore the uses of data science in different industries and check out our profiles of data science careers.

2. Contact Data Scientists in Your Chosen Field

LinkedIn, Twitter, and meet-ups are a great place to start! Ask them what they do during a normal day. Get recommendations on skill sets (e.g. R vs. Python). Talk to them about your education options. You may not need a bootcamp to land a job. A few months with Coursera or other MOOCs and a good textbook could do the trick.

3. Research Requirements in Job Listings

But don’t take them too much to heart. Sometimes, HR departments dump a huge list of requirements into the skills section and hope for the best. When in doubt, ask your mentors what knowledge is critical to have.

4. Ruthlessly Assess Your Skill Level

This will help when it comes to deciding whether you need a part-time course in Python (e.g. General Assembly), a crash course in Hadoop, or a full three-month immersion in major data science technologies (Metis, NYC Data Science Academy, Galvanize, etc.). Think about soft skills too. Do you need practice developing your own projects and public speaking? Do you need some experience leading a team?

5. Contact Bootcamp Graduates for an Honest Opinion

Many bootcamp organizations list their alumni on the website, but you can also do a LinkedIn and Twitter search. Ask for an inside take on coursework, instructors, job preparation and career support after graduation.

6. Create a Budget

That $16,000 fee is just the beginning; you’ll also need to think about food, transport, lost wages and accommodation. In a place like San Francisco or New York City, housing can be very expensive. Some bootcamps offer merit, need-based scholarships, and scholarships for women; be sure to ask about options.

7. Draw Up a Shortlist

You can start with our Mega List of Data Science Bootcamps, but you may also want to do your own research. What kind of reputations do the instructors have? Can you commit to full-time? Some bootcamps are highly selective – what are your realistic chances of getting in?

Graduate Degrees vs. Bootcamps

Let’s say you have a bachelor’s degree in a quantitative science (or a related field) and you’re thinking of becoming a data scientist. You’ve done a few online courses (e.g. Coursera, Udemy, etc.) and are ready to invest in more education. You’re considering three options – data science bootcamp, data science masters, or PhD in data science. Which one do you choose?

We don’t have the definitive answer. Unlike, say, medicine, there is no tried and true path to a career in data science. Some gurus hold a PhD in statistics and have built up an arsenal of data tools; others only have a B.S. and an incredible portfolio of projects. Hard-hitting entrepreneurs have created start-ups after minimal time in an academic setting.

Interview with DDL Graduate

Mehdi El-Amine

Bootcamp: District Data Labs – Data Science Incubator
Title: Data Analytics Manager
Company: www.circleback.com

“You can take all the online courses in the world – you can buy books, read blogs, do code exercises – but eventually you need to get your hands dirty with a real project.”

Why did you choose to apply to District Data Labs? What influenced your decision?

Location was a factor – the DC Metro area is not as rich in data science education offerings as, let’s say, the Bay Area or NYC. Also, the incubator was free of charge; you just had to submit an application and get selected. The format (mostly group self-directed) was a great fit for my schedule as a full-time professional, and it offered me the chance to do hands-on data science work outside of the paradigm of “class-based education.”

What kind of data science skills and experience did you have before you started?

I have a Bachelor’s in Computer Engineering. At the beginning of my career, I was a full stack .NET guy, with good knowledge of SQL databases. I’d done some amateur web development with the LAMP stack, too. Around the late 2000s, I did some cool stuff with OLAP databases (or, at least, it was cool in my head). So I did have a technical foundation, albeit a dated one.

However, in the few years leading up to the DDL Incubator program, I was far removed from all kinds of technical work. Eventually, I quit my job and some old professional connections helped me land a new gig as a Data Analyst. I had just been introduced to Python and I started using it at work, along with a lot of SQL and some exploratory data analysis work (the single-machine-csv-type, not the multi-node-cluster-hadoop type). From there, I put my hands into every data-related project I could find, at work or outside of it.

What goals did you have for the bootcamp?

Most of all, I wanted to do some hands-on data science in the “wild.” You can take all the online courses in the world – you can buy books, read blogs, do code exercises – but eventually you need to get your hands dirty with a real project or your learning will be incomplete.

Deep down, I knew this. While I was taking online courses, I would think to myself: “this thing I’m working on is very directed; if someone were to come to me and hand me the problem statement with no additional instructions, I wouldn’t know where to start, where to get data, what modeling technique to use, etc.”

What was the application process like? Were any parts difficult?

You would go to Tony’s office (Tony Ojeda is the DDL founder), he’d give you one ingredient, and you had to make a meal out of it in under 20 minutes. Then you serve the meal in three plates, and you juggle the three plates while riding a bicycle.

Okay, maybe that’s not entirely accurate. You don’t have to ride a bicycle. The application process is actually a form with some questions – some more elaborate than others – but I don’t recall anything being especially difficult.

What was the coursework like?

The incubator program did not involve much coursework, if any. There were one or two tutorial-like sessions at the very beginning on the expectations of the program – how to report on progress, the schedule of deliverables, and dealing with the github repo. And there was instruction on how to build a data product.

But the act of skill-building was mostly your own responsibility; you could reach out to staff for guidance or help whenever you needed it. I ended up learning a ton of new skills, a lot of things that were intangible – yet crucial – to managing and executing a data project.

Did you work on a portfolio or capstone project? What was it?

The program was project-based. A few sponsoring organizations came in Week 1 and presented projects that we could work on. We were split in teams and each team would choose one of the projects, or come up with their own.

If you did work on a sponsored project, you were able to collaborate – to some degree – with folks at the organization itself (obtaining data, understanding the knowledge domain, getting direction on which questions to try to answer, etc.).

My team and I chose to work on a project by the Bureau of Labor Statistics. They proposed some work on identifying geographical regions with significant demand for (or supply of) particular job skills. But I decided that we should do something bigger.

Mind you, I had no data science skills, but that did not deter my ego; I somehow convinced my team that we could figure out how to forecast national employment growth, using BLS historical labor data. The idea was to take some economic variables such as the employment-population ratio, or labor participation rate, throw them into a multi-variate regression and forecast the monthly unemployment rate.

How would you judge the quality of the instructors and/or mentors?

Excellent. Benjamin Bengfort was the master data scientist with all the answers to both low-level questions (what library to use? how to scale this or that, etc.) and high-level ones (Is linear regression the right approach? are these findings useful?).

Laura Lorenz helped us engineer better code, structure it, and design a proper data product. She also made sure we stayed on track every two weeks with our deliverables, which was definitely needed!

What kind of career training and job preparation did the bootcamp provide?

Even though the program wasn’t structured as a training program, it still provided a huge amount of learning that was directly applicable to the professional work environment. Executing a data project and putting together a data product is a different beast from the standard exercise of “get some data, get an ML library, get predictions.”

In no particular order, you need to:

• Figure out the right questions to ask
• Break the work down into tasks
• Schedule and manage the tasks between yourself and your teammates
• Automate data retrieval and wrangling
• Test and validate your model(s)
• Deploy your model(s) into a production-type environment
• Build a web front (app, website, api) for others to consume data from your model(s)

We worked on all of these pieces over the course of the project.

How many people were in your cohort? Was there a lot of diversity?

Off the top of my head, the number was somewhere between 30 and 35 people, split roughly 60% male and 40% female. The folks in my cohort had different, but complementary, professional skills; so many of the projects presented at the end of the incubator were damn impressive! My own team was composed of an African American, an Indian, and a Middle Eastern Arab.

Were you guaranteed a job after graduation?

No, but I heard one or two people were planning to change jobs after the bootcamp. I wouldn’t be surprised if they did.

What are you up to now? Do you feel the bootcamp helped get you to this stage?

I am still with CircleBack, working on new and cool data projects. The bootcamp helped me almost immediately; I now make use of several tools I picked up (IPython notebook, pandas, matplotlib). I evaluate the performance and accuracy of data models, which I did not know how to do before the bootcamp. And I also interact with a lot of data APIs (our own or third-party vendors), which is something I got better at after working on the BLS project.

Would you recommend DDL? Who would be the best candidates for it?

Absolutely. The incubator itself is invaluable, but plenty of DDL’s standard workshop-type course offerings are also highly educational, too. If you were to apply to the incubator program, I would say you need:

1. A motivation to learn
2. One technical skill, which could be anything (SQL, or R, or statistics)

Do you have any other advice for folks interested in a data science bootcamp?

Don’t rely on being fed the information; you have to want to figure it out yourself. It doesn’t matter what the tool is – Python, R, or other – the most important skills you’ll learn are centered around data science “thinking.” Is this a good training set? Am I overfitting? How would I interpret the predictions of this model?

Finally, choose to work on something small and narrow-focused; otherwise, you won’t be able to get all the data you need, or figure out which features are relevant, or have useful results. The big questions can only be answered after the small questions.