Data Science Degrees: who should be taking them?

With all the hype and excitement around Data Science over the past few years, a significant proportion of the worlds' universities have got on board and are now providing Data Science Masters and PhD degrees. At first glance this may appear to be a good thing. After all, Data Science is now being recognised as a profession in its own right. Seizing the opportunity, the higher education sector has decided to jump on board in an attempt to fill the supposed education gap this new profession provides.

I suppose it would be a good thing if the majority of those completing these degrees and certificates were able to use them effectively. What I've noted, however, is that although some graduates slot perfectly well into the commercial world and are able to leverage their education appropriately, this is not the norm. A large proportion have difficulty - some more than others. 

The problem with data science is that, irrespective of the sub-speciality, be it image analysis, financial predictions or race forecasting, a data scientist is only as good as their life experience. There is an additional non-academic learning curve required for graduates, and that is to be able to adequately understand relevant aspects of the field in which they work. This takes time. Unfortunately throwing a newly minted graduate straight into a machine learning frenzy often does more harm than good. More than a few data science projects have failed for just this reason. Knowing how to use the tools of the trade is one thing, but knowledge of the business is paramount.

The only way to gain this knowledge is through experience, which is why the best data scientists have a history; they've had a previous life as an analyst, dashboard developer or back end programmer. Their foray into data science was built on top of this background, which has been essential to their maturation into, and success within the role.  

So whilst I commend the university sector for their recognition of the data science profession, I think more thought needs to go into who exactly should be taking the courses. I envisage that more businesses will in the future form links with the education sector, and that existing employees that are regarded as suitable for the leap into the world of data science will be sponsored by their employers to do so. Data science should be regarded as a 'next-step speciality' role, rather than a primary one. 

Perhaps I am being too deterministic here, but I think more efficiencies will come out of this approach, and given the large number of data science project failures, there is nothing to lose by going down this path. It will most likely empower those most likely to be able to leverage the power of data science - those who really know the business.