An Automated Allocation Innovation

Leader:
• Kiran Gajwani
Team:
• Danielle Doyle
• Jeffrey Miron

Project Overview:
Across the University, we face situations where the fundamental challenge is allocating individuals to a limited number of slots based on individuals’ preferences, seniority, prerequisites, concentrations, affiliations, and so on. Our innovation is to create an automated, efficient, and flexible computer tool for these allocation challenges.

Examples of allocation challenges at Harvard are: allocating TFs based on TF and instructor preferences; assigning students to one course in a series, when one can’t use the Sectioning tool; creating course-specific enrollment lotteries for oversubscribed courses; assigning work to committees based on committee members’ preferences; assigning individuals to groups or choosing individuals for slot-constrained events, based on preferences or other attributes; and more.

Project Objectives:
Our goal is to create a user-friendly, freely available, web-based allocation tool that will efficiently and effectively handle the large variety of allocation problems that exist across the University. Our tool will be flexible to the user’s choice of allocation method for their particular problem.

Optimal allocation problems are complex and depend on how one thinks about individuals’ well-being. Thus, we will use research on allocation algorithms and welfare maximization to create a tool that will produce fair outcomes while saving time and eliminating errors.

Expected Outcomes:
The tool will exist as a standalone web-based interface, where users can either upload a spreadsheet of information or use the tool itself to gather preference and other data on the relevant individuals. The user will choose the attributes that they want considered in the allocation algorithm, in whatever priority they prefer. The user will also select how they want the algorithm to consider individuals’ well-being. Additionally, users will be able to quickly calculate and visualize data and statistics on the allocation results. The tool will also be designed in a way that is amenable to integration with other systems.

Expected Impact:
Given the variety of allocation issues that we know exist at Harvard, we believe staff, administrators, faculty, and students across the University’s schools and administrative units will use this tool and reap large efficiency gains from it. To the best of our knowledge, there is not only no such tool with these features at Harvard, but no available tool anywhere.