Top Master of Engineering project teams and students are recognized at graduation ceremony
Team-based projects working on real problems for a diverse set of clients constitute the capstone educational experience within ORIE’s Master of Engineering (M.Eng.) program. At the 2016 Graduation Ceremony, 5 project teams were singled out for special recognition through a prize endowed by the Silent Hoist and Crane Company more than a half-century ago. The prize recognizes “aptitude and special interest in the Art and Science of Materials Handling,” which “has come to include the handling of data and information, recognized as vital operational functions in almost all organizations” according to M.Eng. Director Kathryn Caggiano.
Selecting the most promising products for enhanced presentation on a retail web site
First prize in this year’s competition, which is open to all students and teams in the College of Engineering, went to a team of ORIE M.Eng. students that devised a way to decide which products should receive enhanced presentation on The Home Depot’s web site. With a very large number of products displayed online but limited resources, The Home Depot needed a way to figure out which products would benefit most if accompanied online by one or more resource-intensive display enhancements, such as including videos or 360-degree images. Many products on the website have already been provided with display enhancements, so the team was able to attribute some “conversions” (from “just looking” to “I’ll buy that”) to these enhancements. The team developed a software tool that enables company staff to predict, for a specific product, how different enhancements would impact overall conversion, allowing The Home Depot to optimize the set of items chosen to receive new enhancements and thus increase overall online sales.
Visiting Assistant Professor Frans Schalekamp and Assistant Professor Siddhartha Banerjee advised the team, which consisted of students Kaiwen Lv, Dongwei (Jackie) Liu, Shengjie Tu and Xueqi Zhao, seen at left flanked by Caggiano and Banerjee. Home Depot’s Manager of Analytics, Jenny Tate, commended the team and advisors for their work, saying that “we look forward to implementing the solution this summer.”
Reducing conflicts in the schedules for evening prelim exams
For many undergraduates, fall and spring evening preliminary exams (“prelims”) can impose additional stress if two exams are scheduled for the same Tuesday or Thursday night. While professors can make special arrangements to deal with schedule conflicts, doing so costs extra time and effort for professors and teaching assistants.
The Cornell Educational Policy Committee (EPC), which oversees exam scheduling, asked an ORIE M.Eng. project team, advised by Professor David Shmoys, to devise a data-driven approach to reducing conflicts. To do so, the team obtained data relating prelim schedules for Fall 2014 and Spring 2015 to student schedules in a way that preserved confidentiality. They found that 1478 conflicts had occurred that fall and 1203 that spring.
Prelim scheduling faces difficulties not found in end-of-semester final exam scheduling, which was previously a subject for ORIE research. Prelims should occur at appropriate times during the progress of the course, and should be properly spaced during the semester. The schedules for the 2014-15 academic year reflect that as far as possible given the process by which they were created (EPC Chair Professor David Delchamps noted that “Cornell's evening-prelim scheduling, for as long as anyone can remember, has consisted of thirty people in a room horse-trading”). But would optimally scheduling some course prelims one exam slot earlier or later have created a better schedule? The team used the computational technique of integer programming to find out. The schedules they devised this way would have reduced the number of conflicts each semester by more than two-thirds, to fewer than 400.
With their work on prelim scheduling, the team consisting of Mingjie Lyu, Ryan Walker, Yi He, and Jingzhuo Xu won second place in the Silent Hoist and Crane competition. Lyu and Walker are shown at right with Caggiano and Shmoys. In light of the team’s results, Delchamps noted that, as EPC Chair,”I often encounter circumstances that cry out for more analytical thinking and data-driven problem-solving than have been customary in the past. I'm happy to steer more problems arising from such situations to ORIE M.Eng. students as they arise.” A team of undergraduates is currently working to enable continuous improvement in prelim exam scheduling with software that uses the team’s approach with the prior year’s schedule and current enrollment data to similarly minimize schedule conflicts.
“The project process was a valuable experience in applying the M.Eng. curriculum to an interesting problem in a way that can have a meaningful impact for students at Cornell,” said team member Ryan Walker. “I am always surprised at how much better computers are at scheduling when compared to humans,” he said. However team member Yi He said he learned from the project that there are limitations to computing power: The first approach the team tried “could theoretically solve the problem perfectly; however it took forever to run. We were stuck for quite a while until we figured out [a way] to reduce the number of available options,” which “taught us a lesson about tradeoffs.”
Walker, from Memphis, Tennessee, came to Cornell with degrees in Industrial & Systems Engineering and Polymer and Fiber Engineering from Auburn University in Alabama. He now works for Green Mountain Technology, a parcel audit and consulting company, where he works in “analytics and network optimization supporting many of the country’s top parcel shippers.” Yi He spent his first 18 years in a small city near Shanghai, China, and completed his undergraduate degree in Cornell’s College of Arts and Sciences.
Enhancing a system that schedules non-urgent air ambulances
For nearly a decade, ORIE M.Eng. project teams have worked with Ornge, an air ambulance and ground transport service for the province of Ontario, Canada, to develop and enhance a computer-based flight scheduling system that is now in regular use by Ornge. During that time, the circumstances in which the scheduling system operates have changed in at least two significant ways: jet planes, which have different runway requirements than propeller-driven aircraft, have been added to the Ornge fleet, and one Ontario airport receives so many requests for service that the local ambulance crews became overwhelmed.
Under the guidance of Professor Shane Henderson and PhD student Patrick Steele, the 2016 M.Eng. project team of Xiaofan (Tracy) Chen, Kathryn Dear, Junhong (Jerry) Guo and Zhengyu Long won third prize in the Silent Hoist and Crane completion for their work to modify the software system to handle these constraints. Chen and Dear are shown at left, with Caggiano. Running the modified software against data for 198 recent days showed that it can yield an optimal schedule in an acceptable time on more than 95% of the days. Although the new constraints actually reduce the number of feasible schedule choices, they introduced the potential for an exponential increase in computer run time, which made it unlikely that Ornge would adopt the new approach. However, Professor Henderson reports that “Jerry continued working on the problem in the summer following the completion of the M.Eng. project in May, and through a series of innovations managed to actually speed up the runtimes compared to the previous unmodified version.” Ornge is now testing the new software prior to adopting it for day-to-day use.
Tracy Chen said that the project gave her “a chance to see how the advanced knowledge that we have learned is used in actual business. The mindset of how this problem has been broken down and formulated under reasonable assumptions has definitely enriched my problem solving skills at a high level.”
Project clients Dr. Russell MacDonald and Mahvareh Ahghari, respectively Ornge Medical Director and Research Coordinator, wrote that “the problems were complicated and complex to solve, yet the team achieved very strong results and a working solution. The team was well prepared and conducted themselves very favorably / positively, producing an excellent result. We possess tremendous opportunities for students to address real-world problems (greatly benefiting them in their training). In return, we derive benefit in solving significant systems-based problems that achieve better delivery of care to our population.”
Two teams win honorable mention
A project for a frequent M.Eng. project client, Cayuga Medical Center (CMC) in Ithaca, and one for a large California wine distributor were each awarded honorable mention in the competition.
The CMC project used simulation techniques to explore the impact of alternative work schedules and policy changes on Emergency Department (ED) patient flow and recommend ways to improve patient satisfaction and effective use of hospital resources. The team, consisting of Jonathan Boroumand, Vishesh Shah, Aradhya Agarwal, (at left flanked by Caggiano, Muckstadt and Pender) as well as Mark Zito, concluded that some simple changes in scheduling have the potential to increase the capacity of the ED by 16% while reducing the time spent in the ED by 16% per patient, at no increase in cost. Assistant Professor Jamol Pender and Emeritus Professor Jack Muckstadt advised the team.
Shah, who grew up in Mumbai and earned an Information Technology degree there, said “the project was the best learning experience in the M.Eng. program,” but “without the stuff we learned in the fall, we wouldn’t be able to get the results we did.” The project was “the first time when we weren’t given the question but still had to work on finding an answer.” He is currently on the marketing team of supplyhouse.com.
The wine distribution project team (at right), consisting of Anavi Nahar, Gautam Nichlani, Suraj Naghusbana Ponnaganti, and Henry Dong (not shown), developed an Excel application that can be used to optimally rebalance inventories across Regional Distribution Centers (RDCs) whenever a shortage at a particular RDC is anticipated, thereby avoiding the need for expensive expedited shipment by truck directly from the central warehouse to the retail store. The team also devised a program, using historical data, that simulates demand across the RDCs. They predict that rebalancing inventories can reduce transportation costs by between 20 and 25%. ORIE Professor David Shmoys and Visiting Assistant Professor Frans Schalekamp advised the team.
Wine distribution team member Suraj Nagabhushana Ponnaganti noted several general lessons he learned from working on this project: “Don’t dive into building a solution until you have considered all aspects of the problem. Understand the problem statement and define the scope and objectives of the program (turns out, building a solution is not the toughest part!) Quit searching for an optimal solution when you reach your tolerance limit – any solution better than the current process with be appreciated!”. A graduate of Fr. C. Rodrigues Institute of Technology in a planned suburb of Mumbai, where he grew up, he was hired by the M. Eng. project client in California and is participating in a structured program for their new recruits.
Team member Anavi Nahar found the project “beneficial to develop certain non-technical skills, essential for my current rule, through interaction with my team, advisors, and a client in industry.” She grew up in India but came to the US five years ago. She studied mathematics and statistics at Purdue University before coming to Cornell for her M.Eng., and now works in Seattle as a Program Manager in Azure, Microsoft’s cloud platform. “I would also like to help make the kind of impact Cornell has had on me to future ORIE M.Eng. students,” she said.