It’s been a couple of weeks since Johns Hopkins issued final certificates for their Data Science Specialization on Coursera. I’m glad to say that I am now among the first crop of “alums” of the program. According to the last email we students received from our Johns Hopkins professors, about 2.3 million students have attempted at least one of the courses in the Data Science Specialization. Of those, 68,000 verified certificates were issued for completing a single course. 800 students made it through all 9 courses and enrolled in the capstone, and 500 of us successfully completed the capstone.
The JHU Data Science Specialization has been a very interesting, intellectually stimulating experience. I’d never even taken a single MOOC (Massive Online Open Course) before I enrolled in the first course for the specialization. The courses were a combination of video on demand, computer graded quizzes, and peer graded projects. Given the huge enrollment numbers, there’s obviously a lot of interest in the program, and I continue to get asked questions about it on LinkedIn. I’ve written about my experiences with the specialization before, including a review of every class, but now that I’ve completely finished it and started working in industry, I’d like to give a high level review of the data science specialization.
Strengths of the Data Science Specialization
There’s plenty to talk about here, as the program offers a lot to those that throw themselves into the challenge.
- It’s possible to start with very little data science knowledge. You will be at a noticeable disadvantage if you don’t have at least a beginner’s knowledge of computer programming or statistics/undergraduate mathematics. However, there is no need to know a random forest from a random uniform distribution. The Johns Hopkins professors don’t assume any data science background of any type.
- The Data Science Specialization uses R, a very powerful and popular language for statistics/data science. I don’t see how someone could make it through the JHU program gaining at least an intermediate working understanding of R. R is free and open source, and as a result there is a huge community supporting it with packages and tools. Those tools are emphasized throughout the program, and you’ll have the opportunity to share your work on GitHub.
- A broad variety of data science topics are covered. While the specialization starts slowly with courses like R Programming, Machine Learning doesn’t come until the 8th course. That’s because there are earlier courses emphasize statistical inference, reproducible research, exploratory data analysis, etc. Even within each course, there are a wide variety of topics covered.
- Every course had at least one programming based project. This is where the real substance of learning happens in the Data Science Specialization. I found that there was no way to work through these projects without gaining significant understanding as they were challenging and open ended.
- The program is affordable. The entire specialization can be completed for $500 (just $50 per course). A 1 year master’s degree program could cost $50,000 at many universities. I don’t think that the Johns Hopkins specialization is as valuable as a master’s degree, but I think it has far more than 1% the value of a master’s degree.
Weaknesses of the Data Science Specialization
For everything I did like about the specialization, it is not without its weaknesses.
- Because very little prerequisite data science knowledge is required for the program, the instructors can’t go as deep into content as they could with if they had required students to have a deep knowledge of statistics or computer science up front, like many on campus programs probably do. I don’t think I gained mastery or depth of certain topics that now seem important to me, particularly with topics that occurred later in the sequence like machine learning. That said, I was one of the students who benefited from the slower pace at the beginning of the sequence, when I was struggling to keep up, so it’s hard to complain about this very much.
- Although I really appreciated going into great depth with an industry standard tool like R, I do wish there had been at least minimal exposure to other tools. In my opinion, it does weaken the Data Science Specialization somewhat to offer no exposure at all to Python. I’ve been left to try to figure out my own introduction to doing data science in Python with Pandas, as it is a valued skill in industry, but with fewer options for online training than R. There was also a no real introduction to “big data” with Hadoop tool sets, or even a discussion of how a Hadoop cluster works. Even a guest lecture on this topic would have been beneficial.
- Finally, the JHU Data Science Specialization has something in common with Coursera and all MOOCs–learning on your own is a challenge.Yes, there are forums where you can discuss problems with other students and community teaching assistants. But while they are of significant valuable in the learning process, course forums are not a full replacement for having an in-person conversation with a classmate, colleague, or professor about a sophisticated topic.
Bottom Line Data Science Specialization Review
The Data Science Specialization from Johns Hopkins is a terrific learning experience if expectations are set properly. If you are coming from a field that is only tangentially related to data science as I was, this program can expose you to exciting new ways of thinking about solving problems with data. I am certain I would not have been able to embark on my new career path so quickly without having gone through this sequence of courses. My Johns Hopkins professors got me up to speed on many data science topics in a very short amount of time, and I’m incredibly grateful. On the other hand, I think there is a ceiling as to how much one can learn with this type of program without practical experience in the field, and the ceiling is lower here than with a full blown degree.
Don’t take that as a significant criticism of the sequence or Coursera, as I’ve recommended both to many people and will continue to do so. It’s just a bit of advice to set your expectations realistically, given what a vast and complicated world data science is.
My reviews for the rest of the Data Science Specialization follow:
Course 1: The Data Scientist’s Toolbox
Course 2: R Programming
Course 3: Getting and Cleaning Data
Course 4: Exploratory Data Analysis
Course 5: Reproducible Research
Course 6: Statistical Inference
Course 7: Regression Models
Course 8: Practical Machine Learning
Course 9: Developing Data Products
Thoughts on Completing the 9 Johns Hopkins Data Science Courses