What Makes Data Analytics so Complex?

Jana Schaich-Borg, who co-teaches a popular suite of online courses on data analytics, discusses the fast-moving field.

Jana-Schaich-Borg.jpg
Jana Schaich-Borg

Three questions on data analytics with Jana Schaich-Borg, who co-developed and co-teaches Duke’s suite of online data science courses hosted by Coursera. (Read more about those courses here)

Is there something specific about data now that makes it so complex?

Kind of. It's easier and cheaper than ever for companies to store unstructured data.  To learn something about these data, somebody has to learn how to organize them and turn them into a format that can be analyzed.  Such data merging and wrangling is taking a constantly increasing proportion of analysts' time.   Nonetheless, the analysis that comes after all this data wrangling is just as important, if not more important, as it was when a larger proportion of collected data was highly structured.

Is there a single key frustration or challenge you’re hearing from your students?

Yes, they are very frustrated with the lack of opportunities to learn how to work with real company data.  As a consequence, I made it a priority in my courses to make sure students would be able to learn the technical skills I was describing through practicing with real corporate data sets that were used to address real business problems.  These data sets gave students a chance to see how messy real data are, and helped them gain experience with techniques they could use to handle such "mess" in their own jobs.  It also helped them see how to use technical skills to answer questions that would provide actionable guidance to a business.

Is there a single key frustration or challenge you’re hearing from companies trying to hire analysts or data scientists?

Yes. Companies constantly report that although their analysts often have the right technical skills, they don't tend to have the necessary communication skills.  This lack of communication acumen prevents their analysts from learning what their stakeholders really need, limits their ability to address meaningful questions that provide value to the company, and inhibits their ability to make sure any value they do create gets realized.  It's clear that you need much more than programming skills to be a competitive data analyst or data scientist.