I don’t have an OR background, at all. I’ve studied computer science, machine learning, cognitive science, psycholinguistics, and statistics, but I’ve never taken a course in optimization. I’ve taught myself a few things about it, over the course of my working life, as well as topics in quantitative marketing and business analytics. Yet somehow I’ve been invited to be a blogger for this conference. I clearly owe you an explanation, if not an apology.
A few years ago, I co-founded a Meetup group called Data Science DC, with the goal of bringing together a wide range of practitioners in fields such as statistics, machine learning, data journalism, business analytics, and yes, OR. The group took off rapidly, and in 2012 several of us formed an umbrella group called Data Community DC. DC2 now consists of six different Meetup groups with nearly 6000 unique members, and runs workshops, social media, and more.
DC2 was a springboard for connections in our regional professional community, including WINFORMS, the DC chapter of INFORMS, where I’m now a board trustee. Two years ago, WINFORMS hosted a panel on OR/MS and the new Analytics, Data Science, and Big Data buzz. As one of the panelists, the title of my talk was “OR is one way to Analytics / Data Science excellence“. I still think that’s true. Here are versions of a couple of my slides from that day:
Current buzzwords (“data science”, “business analytics”) are just like OR.
Mathematical / statistical sophistication is applied to complex (business) systems, using computational tools.
Value comes from combining domain expertise with numbers.
The best practitioners bridge multiple worlds.
Great communication is key.
Sound familiar? This is exactly what I’ve seen in my experiences at INFORMS events (including this conference, several years ago). And this is why INFORMS, the organization, has been so focused on getting its point of view out, with initiatives such as CAP certification, the rebranding of this conference, and the new big data conference.
But there are some cultural differences too.
Current buzzwords depart from traditional OR
more insight, less action — deliverables tend towards predictions and storytelling, versus formal optimization
more openness, less big iron — open source software leads to a low-cost, highly flexible approach
more scruffy, less neat — data science technologies often come from black-box statistical models, vs. domain-based theory
more velocity, smaller projects — a hundred $10K projects beats one $1M project
more science, less engineering — both practitioners and methods have different backgrounds
more hipsters, less suits — stronger connections to the tech industry than to the boardroom
more rockstars, less teams — one person can now (roughly) do everything, in simple cases, for better or worse
Some of these things are huge advances. The impact that a single smart person can have on an organization is vastly more than it was ten years ago. Some of them are arguably steps back. And the value of OR’s participation in the broader analytics community is, I would argue, to bring decades of maturity to the table, and help focus people and organizations on the bottom line — results that matter.
To me, the OR way of approaching organizational problems is particularly strong at asking business-value questions, identifying and then modeling the problem, and then improving and optimizing processes, not just providing predictions and visualizations. Combining those traditional skills with new tools, new fluency with software and programming, and new career paths and business models, should be a winning approach.
Thoughts? Leave a comment, find me in the hallways, or tweet me @harlanh!