Education

After the success of the video tutorials, I began exploring how to impact student learning using videos in other ways. I chose to flip the course ISYE 4220: Optimization Algorithms and Applications. This course is taught within an Industrial and Systems Engineering program as a technical elective, although it draws students from other engineering disciplines and business analytics. The course focuses on the design, analysis, and implementation of algorithms for combinatorial optimization problems. A key learning objective is to foster the students' ability to apply concepts in computer programming and theoretical computer science to the field of OR.

Within the course, students tend to struggle with the abstract nature of pseudocode while designing and analyzing an algorithm for a particular combinatorial optimization problem. In particular, many students are able to understand the `high-level' steps involved with the algorithm but struggled with how each line of the pseudocode translated back to these high-level steps. These connections often become clearer when we could step through a small example to show exactly how each line of the pseudocode would be executed. This understanding typically makes it easier to implement the pseudocode since the students have a better idea about the inner workings of the algorithm. However, under a traditional classroom environment, it was not possible to provide the details of applying the pseudocode to a small example for every algorithm that was analyzed.

In Spring 2016, the course was taught in a manner that flipped the design and analysis aspects of the course. Students were provided short video lectures prior to class (1-2 videos each between 10-15 minutes) that were to be viewed outside of the classroom. These video lectures designed an algorithm, analyzed it, and provided the application of it to a small example. The first part of each class involved students completing homework-like exercises on the material from the video lecture. For many algorithms, the second part of class would then focus on the implementation of the algorithm. The class concluded with a `challenge' to the students that introduced the concepts presented in the next video lecture. The purpose of the challenge problem was to allow students to begin gaining intuition about the algorithm (or concept) that would be introduced in the next video lecture. Student achievement, especially as it pertains to the coding projects, increased due to this flipped classroom environment.

I am a firm believer in using projects to demonstrate the important philosophy of "all models are wrong, some are useful" (attributed to George Box) within an OR classroom. I hope this helps students understand that it is as important to justify the assumptions used to build their models as it is in solving the model correctly. I have created a number of "toy" projects but also had the opportunity to create a project by working closely with collaborators from Deloitte. You can read the case article and access the case materials through INFORMS Transactions on Education.