The Business of Moving People

Not infrequently, I am asked about my academic specialty. When I reply "operations research", as I usually do, I'm often met with a polite but blank stare. This happened to me a couple of times at a recent party. If the questioner inquires further, I'll try to give an example or two of what OR folks work on in the real world, while omitting any references to "integer programming models" or "Markovian decision processes".

Two recent articles in the popular press do a nice job of illustrating real-world problems to which operations research models and techniques are applied. Coincidentally, they both have to do with getting people from point A to point B, hence the title of the blog post. I thought I would link them here.

The first is an article at CityLab about bike sharing programs: "Balancing Bike-Share Stations Has Become a Serious Scientific Endeavor". Bicycle sharing services are a relatively recent innovation, primarily in urban environments. By allowing individuals to borrow or rent a bicycle (owned by the service) for commutes or to run errands, the services help drive down traffic congestion and air pollution, while presumably turning a profit (at least if privately owned). As with car rental services, bicycle sharing services allow you to pick up a bicycle at one location (station) and return it at another, a necessity if the bicycles are going to be used for commuting. (You do not want the bicycle out of service for eight or nine hours while the renter is at work, and the renter does not want to pay for entire day's use of the bicycle.) Also in common with car rental agencies is that this can result in some locations developing a surplus of available bicycles while others are starved. The CityLab article focuses on the twin problems of predicting where the imbalances will be and efficiently routing vehicles used to rebalance the stations (moving bicycles from overpopulated stations to underpopulated stations). Neither "analytics" nor "operations research" is mentioned in the article, but clearly that is what is going on.

The second article, appearing in the New Republic, has the rather amusing title "Why the $#&@! Did Your Airline Cancel Your Flight Today? They Had a Very Good Reason." The author, Amy Cohn, is an industrial and operations engineering professor at Unmentionable University in Ann Arbor. She gives a very persuasive argument, understandable I think by lay people, that airline flight cancellations are not motivated by concerns that the flight is underbooked and therefore unprofitable, but rather by a complex planning system (more operations research) reacting to problems and trying to find a solution that does as little damage as possible to the overall schedule, in the process hopefully inconveniencing as few passengers as possible. (Yes, I realize I just said something positive about someone from Ann Arbor. As soon as I post this, I will go do some act of charity in contrition.)

Of the two articles, I think I will be more likely to cite the first one when someone asks just what operations researchers do -- not because the second article is less compelling (or because its author has an unfortunate work address), but because associating "operations research" and "flight cancellations" may not have a positive impact on my popularity.

(Hat tip to @or_exchange, @HarlanH, @portlypoor and @Neil_Irwin for tweets leading me to those two articles.)

Update: Another tweet led me to a blog post by Jean-Francois Puget titled "Optimizing Real Time Decision Making". It uses a taxi service to illustrate how OR tools (optimization) can actually improve aggregate system response time for customers by intentionally delaying responses to some service requests (another people-moving example). Elevator dispatch algorithms provide a similar but not identical application (but good luck explaining to someone that there is actually a reason for them to wait for a car while another car sits idle).

(Hat tip to @dualnoise and @menkris for the tweet that led me to J-F's post.)