The Analytics Edge was originally developed on the MIT campus as an elective for MBA students, but the real-world approach of the MITx course is appropriate for students from a wide range of backgrounds. “Our focus is on the application and the story, and how to use it in real life,” says Allison O’Hair, a lecturer at the Sloan School of Management and one of the course’s co-instructors. “To our knowledge, this is a new way of teaching ... the only class here that teaches analytics through applications. We let students get their hands dirty with real data and applications of analytics.”
The current residential version of 15.071 was started in spring 2012 by Dimitris Bertsimas, Boeing Professor of Operations Research and codirector of the Operations Research Center. In 2011, he approached O’Hair, who was then a graduate student, about changing the course by “focusing more on the applications and the story, instead of the method.” O’Hair helped Bertsimas implement these changes, and they began teaching 15.071 the following spring. This year, the course is so popular that all 85 slots were filled and several people were placed on a waiting list.
The popularity of 15.071 is one reason why Bertsimas and O’Hair have decided to teach their analytics course through edX. “We feel like it’s a hot topic,” O’Hair says. “We thought there would be a lot of interest from world students.” O’Hair also wants to try teaching online because she believes it will become more common. “I feel like it’s the future of education,” she says.
O’Hair and Bertsimas will each give half of the lectures, while TAs will do recitations to review the methods. Much of the course’s material will come from a textbook, “The Analytics Edge,” that Bertsimas and O’Hair are currently writing with Bill Pulleyblank, professor of operations research at West Point.
Students in the course will perform most of their computations with R, a special programming language for statistics that is a free, open-sourced software. In addition, students will be using LibreOffice for two weeks to work on optimization problems requiring spreadsheets. LibreOffice is similar to Excel, O’Hair says, but it is also free and open-sourced.
Unlike most MIT courses, which typically have college-level prerequisites, The Analytics Edge can be taken with a knowledge of high-school-level mathematics, so long as students understand concepts such as a mean or standard deviation. Nevertheless, O’Hair emphasized that 15.071x will be “taught at the same rigor as the residential course. The lectures are very similar, and the problems are the same. There is a huge overlap.”
For one of the classes, students will use data from the Framingham Heart Study to learn how the researchers arrived at their results. This topic ties in with O’Hair’s own research interest of applying analytics to healthcare. For her thesis project, O’Hair created an application, LiA (an abbreviation of Lifestyle Analytics), that uses data from individual diabetes patients to determine their nutritional needs. LiA also shows the impact of various foods and types of exercise on patients’ blood glucose levels, and even designs individualized meal and exercise plans.
In lieu of the in-class group presentations given by residential students, edX students will participate in an online competition, for which they will all receive the same data set and problem. The data will be from Show of Hands, an informal polling application that asks users questions about preferences, habits, and general opinions. Students will try to predict users’ happiness ratings, given the answers to 100 different polling questions and some basic demographic information. The competition will be run on Kaggle, a platform for analytics competitions.
Analytics will infuse almost every aspect of the course: O’Hair even plans to use analytics to study the outcomes of The Analytics Edge. “This is a new frontier people are starting to explore,” she says. “We’re definitely hoping to observe how people handle the homework, how they like the competition …. We’re definitely interested in taking data from the course itself and using analytics to improve it.”