Leader: Cementing OR’s place in the data science / data analytics world


I want to set a challenge for the Society for 2022. It won’t be an easy one, and it probably won’t be one that finishes before the year is out. I want us (yes, you too) to make big strides towards making operational research’s place in the data science / analytics landscape crystal clear.

This is not a new challenge; it’s one we’ve seen repeatedly over decades. How does OR fit in with the latest fad or piece of business-speak? (Remember business process re-engineering?) I could look back to see how long we’ve been grappling with the arrival and impact of analytics, but it doesn’t really matter. The landscape has been complicated by the rise to prominence of data science and AI, and the map setting out the boundaries or areas of overlap of all four has been drawn and re-drawn quite a number of times.

Gavin Blackett.jpg

The Society’s Analytics Development Group (ADG) has been in existence for probably ten years; I ought to check more accurately as we could be missing out on an anniversary celebration. Over this period, ADG has had many successes – the establishment of the Analytics Network and the Annual Analytics Summit being just two of them – but I don’t think anyone who has been involved with the group over this period would claim complete victory in the battle to align OR and analytics. The group’s remit has been extended in recent times to encompass data science and AI but its raison d’etre still remains.

I can identify at least four areas where we need to not only continue what we’ve been doing, but step things up to the next level of involvement and engagement. I’ll tackle these four areas in the coming paragraphs, but without too much thought as to their order or priority.

The simplest one to explain is communication; we need to say far more about how OR shares common ground with analytics / data science / AI. Some of that is fairly basic stuff which we’ve danced around over the years. We need to set out our position on our website for all to see, and to illustrate it with examples. None of this is rocket science, although actually coming to a common understanding of that position might require a few debates. This way anyone, probably someone from outside the profession, can read the position statements and understand our views quickly. Next, we should take every opportunity to share our view with related communities – tell them how we connect, and how bringing OR into their worldview can bring benefits. We have done some of this when opportunities have arisen, but we need to do it again and again and again.

The next is to continue to be involved in initiatives that put us at the heart of these emerging trends. (I wasn’t sure trends was the right word to use – Areas? Subjects? Fields? But you probably get my meaning.) One such initiative is the Alliance for Data Science Professionals (AfDSP). I hope you’ve heard about this already – I’ve given presentations about it to a variety of conferences, meetings and groups, and announcements have appeared in InsideOR. Essentially, the Society is working with other learned societies and bodies to set up professional standards for those working in data science. We have advanced plans to certify the first cohort of data scientists by mid-2022. The details of this work can be shared elsewhere, but it’s being at the heart of such schemes that is important. In the context of mapping the landscape, the work of the alliance has been eyeopening for me; data science really does cover a wide spectrum of jobs, skills and specialities, and of course, that’s one of the challenges OR faces.

Ethical or responsible approaches to the use of algorithms is another key theme that OR can build on, and another area for the Society and the wider OR community to set out its stance. There are seemingly countless examples where algorithms and data handling have caused significant issues. The Society can do its bit to set out how these issues should be tackled and avoided. ADG are working towards sharing a suitable starting point that the community might get behind and develop. Have a look at Richard Vidgen’s paper on this at: https://survivai.ai/resources-2/

I was going to say that the final area was somewhat of a personal crusade of mine, but the truth is that I haven’t been able to do much crusading (yet?). I think the OR community needs to do more to sell what it is we do and how we go about it. We need to shout about ‘the OR methodology’.

Our approach to all aspects of problem solving is one that should be adopted wherever data science, analytics or AI is undertaken. It can bring a vast number of benefits to any analytical work. To put it simply, we need to productise ‘the OR methodology’. We need to have guides, training courses, checklists and all manner of tools to make it simple for ‘outsiders’ (forgive the term) to join our community, to recognise the advantages of our approach and to apply the principles easily in their work.

So, what do you think? Is this a challenge worthy of our attention? Is it a challenge we can meet? If you’re interested in helping, get in touch!