What Is Data Science Specialization, and What Can It Do for You?

Steve Pierson, ASA Director of Science Policy

Previous issues of Amstat News have highlighted new bachelor’s, master’s, and doctoral programs in data science and analytics. Acquiring data science and analytics skills and certifications is also possible through massive open online courses (MOOCs). A leading MOOC specialization was started by Brian Caffo, Roger Peng, and Jeff Leek of The Johns Hopkins University. Here, they provide answers to a few questions we posed to them.

Brian Caffo is a professor and director of the graduate programs in biostatistics. He co-created and co-directs the Data Science Specialization MOOC.

Roger Peng is a professor in the department of biostatistics. He co-directs the Data Science Specialization MOOC.

Jeff Leek is an associate professor in the department of biostatistics. He also co-directs the Data Science Specialization MOOC.

The Johns Hopkins University

Please describe the basic elements of your data science/analytics curriculum and how the curriculum was developed.
Broadly, the curriculum was developed to mirror necessary steps in a data science process. The program starts with an overview and a course in R programming. The R programming language factors heavily into our program, and all subsequent courses depend on it. Students next take courses on getting and cleaning data, exploratory data analysis, and reproducible report writing. The core statistics portion of the program includes courses in statistical inference and regression models. Then, there is a course in machine learning, focusing on implementation and concepts over technical and mathematical details. The program finishes with a course on developing data products, where R Studio’s Shiny framework for developing R web apps is a focus. Students who pass the curriculum can then take a capstone project class, where they put everything they’ve learned from the courses together.

What was your primary motivation(s) for developing a data science/analytics specialization program? What’s been the reaction from students so far?
We are large believers in expanding educational opportunities and increasing access. The model of having content being free or low cost was appealing to us. We were especially interested in data science, since that is where our interests lie.

How do you view the relationship between statistics and data science/analytics?
Statistics is a subset of data science, as data science includes aspects of data collection, curation, and manipulation not generally considered part of the statistics field. Going by our program, four of nine equal-length courses (i.e., exploratory data analysis, statistical inference, regression models, practical machine learning) focus on statistics, while the remaining focus on aspects of data science, programing, obtaining and cleaning data, reproducibility and report writing, and building data products.

What types of jobs are you preparing your graduates for?
Our program is not a degree-granting program. As a secondary credential, for a student with an existing quantitative background, they will be able to work anywhere that emphasizes the data analysis aspects of data science. For many, our program will be their first introduction to data science and entrance into the field. For others, our program helps them retool to expand their current role at their organization. A common example is a data engineer becoming more involved in data analysis.

What advice do you have for students considering a data science/analytics specialization program?
For MOOC programs, we highly recommend that students build a portfolio. This is essential, as most MOOC programs do not offer a formal degree. A portfolio of code and projects can fill in the gap for lack of a degree and demonstrate mastery of the material.

Describe the employer demand for your students/graduates.
We have seen employers allow our program to serve in lieu of formal master’s programs. In addition, we have seen a great number of employers use our program to retrain their workforce.

What should an employer look for when they see a certification on a résumé?
For our program, and likely data science in general, they should look at the applicant’s GitHub page. They should see interesting project and code contributions.

Can someone with a data science certification walk out of the course and become a data scientist?
It depends on the person’s background. For the typical student from the program, they could work as a data scientist in an introductory capacity. Our more advanced students who are augmenting existing quantitative credentials, they could take on higher-level data science positions.