O.R. in the Classroom – Get Real!

I have been a Woolseyian for a long time now. For the uninitiated, Gene Woolsey, professor emeritus at the Colorado School of Mines, has been the self-proclaimed “principal gadfly of the OR/MS profession” for many years. While I haven’t attended any INFORMS meetings wearing a safari outfit, don’t have a mustache and haven’t given out any diamond pins (I have three daughters to get through college – and then there will likely be some weddings to fund), I have tried to follow Gene’s teaching on having my students solve real problems. After 23 years of teaching, I am thoroughly convinced that it gives my students an edge when seeking employment and makes them better O.R. practitioners.

To come clean here,I have not really followed Gene’s teachings to the letter of the law. I haven’t forced my students to work as firefighters or move kegs of beer around the warehouse. Gene has always insisted that his students perform the work before making a recommendation about how to improve the process. Frankly, I just don’t have the time and energy to pull that off. With apologies to Gene, I suppose what I have done over the years is really a version I’ll call Woolsey-lite. My version of Woolsey-lite is simply this: get real problems in the classroom! There are a number of ways to accomplish this. I will elaborate on some approaches I have used.

Bring Consulting to the Classroom

I HAVE BENEFITED both financially and as a teacher from the many consulting engagements I have had over the last 20 years. I routinely bring to the classroom data from previous projects (sometimes masked as required by non-disclosure agreements).I keep the data in as raw a form as possible,which usually means a lot of extra,irrelevant data,sometimes in spreadsheets; in a few occasions in flat files so large the first step is to figure out how to read the file.Why raw data? Raw data has errors and missing entries. As one of my former students (now the vice president of financial systems at a major U.S. bank) recently stated:“I cannot emphasize enough how important it was to use actual consulting projects and not simply case studies. The real world does not conform to normal distributions and often has missing or invalid data. Many professors fail to teach this extremely important concept.”

Since most of my consulting is in the area of discrete optimization modeling, my consulting is a good fit for projects in our core MBA course on decision models and especially our Optimization Modeling course for our Master of Science program. I have had students work on a variety of problems including distribution system design, truck carrier bid selection, product design, global production facility site selection, power generator scheduling and spot market optimization, and nature reserve site selection. In addition to learning about modeling, some of the models developed are not solvable with available software, so the students learn to come up with ways to generate heuristic solutions as needed.

Another former student who works for a major supply chain software provider stated:“This (the use of spreadsheets and heuristics to solve problems) has been very useful to me in my career because, in many cases, I have been able to solve problems without implementing expensive optimization software.For example, when I was working on a problem at XXX, some constraints could not be expressed in a linear form. Instead of using a nonlinear optimization solver like YYY,I was able to separate the nonlinear parts as a sub-problem and solve it using a simple heuristic. This was very cost effective in terms of time and money.”

Another approach to getting real in the classroom is to bring current projects right into the classroom.

Bring Tightwad Client Projects to the Classroom

AT THE UNIVERSITY of Cincinnati, I am the director of our Center for Productivity Improvement, which is essentially a consulting center where we use teams of faculty and students to solve problems for industry using analytics. I often get calls that go something like this:

Camm: “Well, yes I am confident we can help you with that problem and our hourly rates are $x for faculty and $(x-y) for students.”

Possible Client: “I really don’t have a budget for this.”

Camm: “Then I assume you are not in a hurry for a solution. We can do it for free if you are willing to make it a class project and are willing to interact with the students to answer questions, provide data, describe the problem and evaluate the students’ solutions.”

Possible Client: “I like the second option. When does the quarter start?” 

These types of projects have all of the benefits to students previously described, plus the added benefit that I haven’t solved the problem before, so we are all learning by doing. Projects my students have undertaken include, for example, scheduling mail pick up at a Fortune 500 investment firm to control float and labor costs, and efficiency analysis of bank branches for a major Midwestern bank (using data envelopment analysis). Note that with the current client approach, the entire class might have to sign a non-disclosure agreement.

Do Pro-bono Projects for Not-for-Profits

WITH EMPHASIS on community and university service, another viable option for real projects is to volunteer to lend a helping hand.

In the last 10 years or so, there has been more and more emphasis on student community service. Project work for notfor-profit firms is an excellent opportunity to get the students working on real problems while benefiting the local community. Most disciplines have used this type of project for experiential learning and I believe we in the O.R. community should be doing more of it to promote the greater good and our profession. Like projects in the Tightwad Client category, these projects have all of the student benefits previously described, plus the students tend to feel good at the end of the project! I put projects for the university in this category. Three examples of projects my students have done in the past are: development of an inventory control policy for a soup kitchen in downtown Cincinnati, scheduling the University of Cincinnati Law School exams so as to minimize student conflicts, and referee assignments for a section of the Academy of Management national meeting.

Finally,when all else fails,make the project itself an assignment!

Make the Students Find a Project

OFTEN, I will require that the students generate their own projects. This works particularly well with part-time students who have a current place of employment. However, I have successfully used it with full-time students as well. My only requirement is that the data be real. If my experience is any indication of what to expect, you will be pleasantly surprised by the projects your students will generate. Below is a list of some recent projects from my part-time Optimization Modeling course:

• Optimization of Workload and Travel Time for Insurance Underwriters
  
  
• Bank Staff Scheduling
  
  
• A 401K Contribution Optimization Model
  
  
• Evaluation of Processing Rate of an Injection Molding Co.
  
  
• Minimizing Overtime in an Automobile Manufacturing Facility
  
  
• Least Cost Formulation for Sausage & Hot Dog Manufacturing
  
  
• Transportation Network Optimization for a Fortune 500 Food Manufacturer
  
  
• Staffing Model with Inventory & Labor Constraints in a Seasonal Business
  
  
• Fleet Selection for a Major Express Mail Carrier
  
  
• Transportation Optimization for a Ball Bearing Manufacturer
  
  
• Staff Scheduling for a Public Golf Course
  
  
• Optimizing a new Detergent Additive to Meet Consumer Preferences
  
  
• Fadal Machining Center Tool Order Optimization
  
  
• Bottle Molding Investment Optimization

A note of caution: You should be prepared to get a fair number of Markowitz models using stock data from the Internet (although I will usually require proposals midway through the course and put a limit on how many projects of a certain type I’ll take).

Regardless of how the real problems are obtained or how they are “solved,”I always require a written report and a presentation. There are only a few rules. The student must recommend a course of action and the math needs to be in an appendix. The students need to be able to convince management of the rationale for their recommendations and, if requested, they also need to defend the model and the analysis.

In my opinion, there is no substitute for solving real problems. In addition to theory, there are so many practical lessons to be learned by solving a real problem, using the real data and if possible, the real client.

Solving real problems also gives students a self-confidence booster shot as they enter the workforce. A final quote from one of my students:“Armed with my new found confidence in solving real problems, I entered the business world with an attitude that I can solve any problem I come across. Entering an environment where it was common to hear people say, ‘We can’t do that,’ it has definitely helped my career to confidently say,‘I can do that, and I will have it for you by next week.’ From a problem-solving perspective, Dr. Camm’s Large Scale Optimization course was much more challenging than 95 percent of the problems I have been faced with in the real world.”

I’ll be contacting this former student to get permission to assign the other 5 percent as class projects.