PhD in Analytics
The issue that most animated the audience at the Meeting of Analytics Program Directors on Saturday came up last during the day-long event, during the session about industry perspectives: the idea of a PhD in Analytics. Because this was part of a panel discussion with industry experts and because analytics is applied to real-life data almost by definition, the impression I got was that such a PhD would be mostly geared toward producing doctoral-level industry practitioners, although a PhD is supposed to produce new knowledge beyond the deliverables of an industry project.
The discussion turned into a very lively exchange because an attendee from a business school mentioned that PhD graduates who go to industry are a fail for his program, and one of the industry panelists, with a little smirk on her face that she probably shouldn’t have had, replied that she’d like to challenge that notion.
Of course that was easy for her to say, since she gets highly-trained PhDs who, as soon as they graduate, lose any interest in publishing the research papers they have in the pipeline and focus on solving their companies’ problems to achieve the non-minor goal of remaining gainfully employed. But from the standpoint of that other faculty member in a business school whose program is tuition-funded, the fact that research papers won’t be able to be revised in a timely fashion and may not have the impact they should have because the lead student left for industry is not a successful outcome, both from the viewpoint of the dean who funded that student and from the viewpoint of the faculty member who needs those research papers to get promoted.
I do love that in the United States, people with PhDs are not limited to teaching jobs in universities but can have their skills recognized and sought after by top companies, and I am particularly proud of my own PhD students’ placement record at companies such as Amazon, IBM, Marriott, JP Morgan and Ernst & Young, but it is a fact that faculty’s research output suffers from their doctoral students going to industry because a lot of research papers are then put in limbo. When that happens, the business programs that funded those students don’t get the best return for their money, and the federal agencies that funded research assistantships for doctoral students through grants don’t get the best return for their money either.
Should we all refuse to let our students graduate until the revision of the last paper has been submitted, even if that means the student has to start working without having her PhD yet? This is a very different situation than that of a Master’s student in analytics, who gets a grade for her capstone project before graduation. Yet, there is no doubt that Master’s programs are becoming too short to completely train students in the full scope of data analytics, from descriptive analytics to predictive analytics to prescriptive analytics, and from small data to big data.
In fact a lesson I learned from another session at the MAPD is that traditional operations research skills such as optimization (prescriptive analytics) risk getting shortchanged because companies understand predictive analytics best and don’t yet understand what they can do with the more advanced prescriptive analytics. If we only teach to what companies want, we won’t produce graduates with the skills they will need to succeed in the workforce 5 years from now when companies start seeing the benefits of prescriptive analytics and decide to leverage those as well. We do have to be careful not to teach reactively to what companies think they want today. As academics, we also have to protect our students’ interests and also our own careers in ensuring the concept of a PhD is not denatured through industry pressures.
So what is the solution? Multiple Master’s in various branches of analytics? A Master’s and a continuing education program? A Master’s and a training program held at the companies themselves for its data scientists? A PhD? Or a more applied degree with a new name that would sit between the Master’s and the PhD and be geared toward industry placement?