Dr. Hardeep Johar, Ph.D.
Senior Lecturer in the Discipline of Industrial Engineering and Operations Research
Dr. Hardeep Johar teaches the lecture-based courses Data Analytics and Machine Learning for Operations Research (~50 undergraduate students) and Analytics on the Cloud (~70 graduate students), and the project-based course Analytics in Practice (~200 graduate students) in Columbia’s Department of Industrial Engineering and Operations Research. With the shift to remote teaching in Spring 2020, Dr. Johar met the moment by setting up structures for remote instruction and collaboration, maximizing class time for student practice, and leveraging local and international industry partnerships. Read on to learn more about what Dr. Johar did in his course, what lessons and experiences he’s carrying forward, and the advice he has for other instructors at Columbia.
Set up structures for remote instruction and collaboration
Streamline the scheduling of Zoom meetings.
For the project-based Analytics in Practice course, transitioning to remote was initially challenging. Since all my students in this course are in the Master of Science in Business Analytics (MSBA) program, I solicit consulting projects from industry partners so they experience developing analytics solutions to real-life business problems, which tend to be heavily data driven and involve a lot of coding. I begin the course with an overview of what students need to do, so while they know the problems and what they are supposed to work on, they don’t yet know how to do so. This models the unstructured process that they’ll face in the real world. They will have to gather the necessary data, process and analyze it using machine learning or operations research techniques, and finally present their results to the relevant stakeholders of their client or company, which ranges from the company’s data science group to 50 company officials.
To support my students, I schedule 3-4 required check-in meetings with each student group over the semester, but they can schedule additional meetings with me if they need more help. Each group also has to meet with their company sponsors at least once a week to report their progress on the project. Prior to the pandemic, the vast majority of these companies were based in New York or maybe New Jersey, so students would regularly visit the companies to talk to their company sponsors. It made students feel they were doing something real and meaningful. But that all stopped suddenly when the pandemic hit—the companies moved to Zoom, we moved to Zoom.
The main disadvantage of relying on Zoom for this course was that it made it harder for students to approach me whenever they have questions. I have an open-door office policy for all my courses. For this course in particular, students would often meet with their groups in Uris Library and other similar spaces on campus, so whenever they ran into a roadblock, they would just walk over to my office and ask for help. When we were all remote, students who had questions would have to email me to set up a Zoom meeting—it wasn’t spontaneous anymore. Every meeting had to be arranged in advance.
To overcome this, I sent everyone an email saying, “I’m blocking these Zoom times for you. Just email me to take a block of time.” Initially, I did this using a Google doc, then I shifted to Calendly, which is a really great tool for scheduling these Zoom meetings. We would also schedule our regular 3-4 required check-in meetings over Zoom to ensure progress on the projects. Students also had to meet over Zoom to have their group discussions, which meant that a lot of the informal interactions were lost.
Oddly enough, when I went back to my open office policy now that classes are in person, I found that students now prefer meeting on Zoom. When I polled my students, ~80% of my project groups did not meet each other in person at all. They said it was easier not needing to be in the same location, which is an advantage.
Leverage technology to facilitate peer evaluation.
For my undergrad Data Analytics and Machine Learning for Operations Research course, students also work in groups on analytical projects. While these projects are not industry-based, students still learn how to find relevant datasets, perform analyses, and present their findings to the class. At the end, each group would present for 3-4 minutes to every other group in a “speed date” format so that they can rate each other’s presentations. There’s a saying in Data Analytics that a simple graph that conveys a meaningful result can often be far more powerful than a complex analysis that doesn’t—the presentation is useful when it conveys something meaningful to the person who has to make a decision based on it. As such, the goal for these presentations was for students to convey that what they’ve done is meaningful. I would handle the technical evaluation of the projects, while students had to evaluate how well each group conveyed their results, whether they raised interesting questions or responded with meaningful answers, and how impressed they were.
This “speed date” format was very challenging to manage using breakout rooms on Zoom, so I asked each group to create a four-minute-long video of their presentation for other groups to review outside of class time and allocate points to. While the videos negated the need to manage Zoom breakout rooms and freed up class time for other learning activities, the group allocation of points presented a different challenge—how to asynchronously facilitate group allocation of points when students were all over the world. I ended up writing a web application that they could use to evaluate each other’s projects and rate them. Students had fun producing the video presentations, and I was very impressed by their creativity and quality, so much so that I’ve decided to continue with it.
Maximize class time for student practice
For my lecture-based courses, the transition to Zoom was a little easier. These are technical, programming-oriented courses with lots of hand-on programming on Jupyter notebooks and no slides during class. Students would code in small informal groups in Zoom breakout rooms so they can collaborate and help each other, and the TAs and I can easily rotate among the breakout rooms to offer help. For example, one of my Python refresher coding exercises asked students to obtain data from Wikipedia in a specific format. The goal is less focused on getting the final answer, and more on having students do the coding themselves rather than listening to me tell them how to do it, which is useful but less effective.
Leverage local and international industry partnerships
Organizing virtual talks with industry experts.
For my project-based Analytics in Practice course, because Zoom made it difficult for students to have the less formalized yet important in-person interactions and conversations with their company sponsors, I began inviting guest speakers from companies like Apple and Google to share their insights about business analytics. These are not technical talks, but rather overviews of how these companies approach business analytics—how they view analytics problems, what data challenges they are working on, or how they foresee the future. Logistically, to get 200 people into a room in Columbia is really hard, but Zoom made it easier to invite guests and organize these talks. I intend to continue with this even though we are no longer constrained to being online, because the MSBA students appreciated learning more about the industry beyond the project that they’re working on.
Soliciting projects from international industry partners.
The other big advantage of Zoom is that about half of my industry-sponsored projects now are from companies that are not in New York. Previously, I had a couple of internationally-sponsored projects out of 40-50 projects, but since everyone has gotten used to working remotely, about half of my projects are from companies based in places like Singapore, Paris, and Hong Kong. These international collaborations have been beneficial for my students. For example, I have students from Singapore who value working on projects sponsored by Singapore companies, because it could lead to future job opportunities there, which is important for them.
Advice for Instructors and the Future of Teaching at Columbia
Find a configuration that works for your course.
My big takeaway is that having some classroom instruction is always useful. When I teach in person, I have lots of students coming up to the dais at the end of class to ask questions, share what they found interesting, and continue the conversation. On Zoom, however, once the class ends, it’s pretty much done. Students seem more reluctant to raise questions or otherwise converse at the end because everyone else in the Zoom space can hear everything they say.
While I don’t think the hybrid teaching model works very well—it’s disconcerting to have a disembodied voice ask me a question, and hard to keep track of the chat in addition to the class conversation—I think a model of half the course being completely remote and the other half being completely in person can work, especially for what I do. For a technical course like mine that has a mix of lectures and problem-solving, I can see recording material for students to watch remotely, and then using the in-person classroom time to build on that material with problem solving, student questions, and perhaps tutorials. Videos have the advantage of allowing students to easily rewatch what they missed, while face-to-face presence is better for addressing questions and spontaneous discussions. When I was teaching on Zoom, I used to miss having my students bustling around asking questions. Now that classes are in person again, it’s back! The bottom line is you need to have both. I think it helps students a lot, so that’s where we want to go.
Leverage your interests and others skill sets.
Always be cognizant of what the most effective way to get something across is, and what would make it interesting for students. Everybody has their own little skill or thing that they know how to do apart from their research area or their disciplinary expertise. I enjoy coding, so I built a web application for my students to bid for the industry-sponsored projects that they want to work on, and another for them to evaluate each other’s video presentations. I also started making Zoom recordings of material that I don’t normally teach in class but is useful for the students, such as how to set up Jupyter notebooks on Google Cloud Dataproc, or Unix basics. Students have found these videos more accessible than other resources I refer them to. Building applications, recording videos—these are things I am able to do, so I do them. Everybody can use their expertise or whatever they find enjoyable to add something valuable for students, and students appreciate it.