The unsustainable lack of analytical skills for a sustainable environment

Developing an analytical model to support decision-making is a really tricky task. I did many of them in order to decide whether it was a good idea to buy an apartment or to keep renting. What differed among them was the level of detail taken into account, which progressed slowly as I learned from my previous mistakes. While most of my models pointed out that it was better to keep money on a savings account and do not lend any instead of buying an apartment for the rent that can be earned (or that I would avoid paying since I had to live somewhere), I often forgot to regard the probability that apartments would appreciate considerably in a near future. Looking back, I see that my renter did an excellent investment by renting to me and waiting for what would come. Nevertheless, that moment is gone and it does not make any sense to bet that real estate in São Paulo will keep rising in price as it did in recent years.

When it comes to a trending topic like sustainability, that is supposed to represent the concern of people with protecting the environment for future generations while exploiting it for their own survival, I have been often told of misconceptions and lost opportunities that made many initiatives on that regard produce the opposite effect. And what is worst: as conceived, sustainability is a great but sometimes overlooked application domain to leverage analytical tools.

For instance, I was told of a workplace where someone had the idea of asking for shelves above the toilet washbasins. Since the washbasins were often wet, people got used to place a paper towel below their toiletries bags. As an experiment, such shelves were placed in half of the toilets of a floor. Long after, people started to ask when those shelves would be placed in the other half, since the experiment seemed to be successful. However, they were informed that someone had to collect data to assess if less paper towels were used in the first half of the toilets or not. As many people did not bother to walk more to use the toilets with shelves, it is very likely that the person undertaking such experiment will be surprised after realizing that the use of paper towels actually increased in that half and decreased in the other half. I hope that the person manages to sum up the use on both halves to realize what happened instead of concluding that shelves increase the use of paper towels.

It is attributed to Ronald Fisher the following quote: “To consult the statistician after an experiment is finished is often merely to ask him to conduct a post mortem examination. He can perhaps say what the experiment died of.”. While it would be interesting to count with professional help, I believe that we would be better off if people working with sustainability were aware of the importance of developing analytical skills to avoid working against their own goals.

An advanced school on handling big data for academic purposes (with full grants for Brazilian and foreign students): São Paulo Advanced School on e-Science

São Paulo state’s research council (FAPESP) is sponsoring an interesting school on e-science, which I would describe to the blog audience as analytics for academic purposes. It is said that they will offer full financial support for 25 Brazilian and 25 foreign students. The school will be held next October in the surroundings of my alma matter, Unicamp.

Further information can be found at: http://www.vision.ime.usp.br/~liu/escibioenergy/

How Analytics makes Operations Research the next big thing

Engineers enjoy laughing at buzzwords that they don’t sell. Despite that, some buzzwords represent important paradigm-shifts. They might not propose any technical novelty but they do contribute to empower our methodologies by valuating the presence of certain skilled specialists in large scale projects. Analytics is one of them: it represents the application of IT to support business decision processes. This post aims at showing that its existence can help leveraging O.R. practice in the industry.

The O.R. filet: there is no such thing as a free lunch!

Who did not wondered about that dream job in which all you have to do is what you do for fun? Suppose that you are an O.R. analyst hired by a company which provides you perfect data and a well-defined problem that you know how to tackle. And that’s not all: they do not underestimate the amount of effort that the project will demand from you and your co-workers. In such a perfect world, you just pick that traveling salesman or bin packing problem with that idealistic instances and expend some time experimenting your favorite techniques until you get satisfied with the results. You would probably finish your work very early and have the rest of the day to share a beer and French fries with your friends at the bar (if that happens in São Paulo).

Back to real life realm: from problem solver to problem finder

Unfortunately, there is a huge gap from being hired until possessing that well-defined problem and that perfect data. That is, if you manage to reach that point. If companies had already all of that figured, they would probably have gone beyond with an in-house approach to their decision problems. In such case, the benefit of an external OR consultant work would be often quite shy. Hence, one must mind that the work is not only about solving an optimization problem but rather helping the company to understand what the problem is and how to collect data to properly solve it.

Some interesting discussions about those issues have been recently raised by a couple of OR professionals called Patricia and Robert Randall on their blog Reflections on Operations Research. They have a blog post about data cleanup and two other posts about understanding what is the right solution for the client’s problem (by the way, I’m waiting for the promised sequel – check the first and the second posts).

And then the O.R. team becomes the Analytics division…

What I exposed before reflects the change that is going on in industry, including my workplace. The O.R. team is no longer called once someone “magically” finds an optimization problem that must be tackled within an IT project. By “magic”, I mean that someone working in a project knew about O.R. by chance and decided to invite an O.R. analyst to check it. Instead of that, new projects are supposed to pass through a preliminary assessment of the need of an O.R. approach. The analytics professional comes into scene to complement the team of software architects, software engineers, data modelers, project managers and stakeholders of any non-ordinary project. The role of that professional is to understand how the system can be used to support business decision-making and define whether statistics, data mining or operations research tools are required to accomplish that. Such assessment avoids that something pass uncaught or misunderstood and, of course, creates lots of interesting opportunities for O.R. professionals both at the assessment and later at the project development phase. As a matter of fact, we have plenty of people ready for the job, as I told last month in a post about O.R. job market in Brazil.

A gain-gain scenario: let’s spread the word about Analytics to empower O.R.!

An Analytics assessment of strategic projects would endorse a broader application of Operations Research, what usually means maximizing profit and reducing costs. Moreover, there is a huge workforce available to the demand that such paradigm-shift would incurs, including me and probably you. So let’s make that happen!

This post is my contribution to the INFORMS’ blog challenge of May: O.R. and Analytics. The INFORMS’ blog challenge consists of a monthly topic about O.R. that is proposed at the INFORMS’ blog. If you happen to write about the topic of the month, send an e-mail to them to get your post mentioned.