I recently caught up with Rafael (Rafa) Irizarry, Professor of Biostatistics with the T.H. Chan School of Public Health at Harvard University to hear his unique perspective as an edX instructor regarding the rising demand for data scientists across most industries. For the past 15 years, Dr. Irizarry’s research has focused on the analysis of genomics data. During this time, he has also has taught several classes, all related to applied statistics. Dr. Irizarry is one of the founders of the Bioconductor Project, an open source and open development software project for the analysis of genomic data. His publications related to these topics have been highly cited and his software implementations widely downloaded.
insideAI News: Why have data science roles become so hard to fill? What is being done to address that problem?
Rafael Irizarry: Data science roles are hard to fill because the demand for data science expertise grew more rapidly than opportunities for people to be educated in how to do data science. At the moment, there simply aren’t enough people trained in computing for data and data analysis. Data science is a new field, with new demands, and the training hasn’t kept up. MOOCs and other online opportunities have become so popular because they were quick to offer a way to learn how to do data science. It can be much quicker to develop something like a MOOC (or a series of MOOCs) than it is to develop a master’s degree program at a university.
Now, universities are starting to develop data science programs for undergraduate and graduate students. There are also many, many online resources being developed for data science – it’s a competitive world on the self-taught side of learning data science.
insideAI News: Big companies such as Accenture Federal are hiring chief data scientists – as this role expands in scope and importance, how should education evolve?
Rafael Irizarry: Data science is a broad term that includes data analysts, database managers, and software engineers that develop pipelines for data analysis. Each of these roles requires different training; it isn’t realistic to have one person that does it all. Universities have courses that teach many of these skills in isolation, but what has been missing are the programs and initiatives that bring the courses together in a cohesive program that can serve the needs of what industry is looking for. Additionally, in universities, many of these knowledge bases are taught in somewhat theoretical ways. Working data scientists in industry need more practical skills. We have been seeking to provide those practical data science skills by teaching with case studies in our Data Science Professional Certificate courses so learners can immediately test out applied, practical data science skills on real data sets.
insideAI News: How do online data science courses and MOOCS in general work to bridge the gap in the data science workforce?
Rafael Irizarry: Because universities didn’t have organized data science educational programs, when the demand for data science kept growing, MOOCs were quick to step in and fill the gaps. Although MOOCs and other online courses can be somewhat superficial, they give you an introduction to the skills needed to do data science.
Another benefit of MOOCs and other online data science courses is that many people going into data science are already out in the workforce. They’re finished with their formal education and aren’t going to go back to school full-time. MOOCs allow learning without needing to go back to a multi-year educational program or to leave your job. These courses allow people to learn new skills and apply them directly to the work they’re doing.
insideAI News: Do these programs involve real-life applications of the skills?
Rafael Irizarry: Yes! Absolutely. One distinction of MOOCs is that there are a lot of real-life applications built into the courses. For example, certain programs have a capstone component, which typically require students to use what they have learned in the program to propose, execute and measure the success of a project the same way they would on the job.
insideAI News: As the future of work continues to develop, how have these shifts impacted online education?
Rafael Irizarry: It’s not yet clear how much the future of work is truly shifting, but online education tends to be more flexible. There’s modularity, so it’s often easier and quicker for MOOCs to be adapted than it is for face-to-face programs to be adapted. New content can be added to or old content can be removed from the MOOC as needs change – modularity can be done with any type of learning- but online education has been more open to adopting it.
Sign up for the free insideAI News newsletter.
