Page 5 - 2020 OIT Progress Report
P. 5

In late spring, President Turner formed a task force and charged it with preparing for a fall unlike any that have come before. The various committees involved in this task force generated several large-scale projects for the University, in which OIT was heavily involved.
The projects shared a few common attributes. They were all focused on solving problems that had never been addressed previously. They required diligent sequencing of actions to ensure academic continuity. They required expertise from across the institution to identify the right path and see the project to completion. In the midst of the pressure and compressed deadlines, codenames for the various projects emerged adding a little touch of humor.
We knew that our classroom facilities had to be adjusted to accommodate the six foot radius per CDC guidelines. What we didn’t know is how the existing course schedule and room assignments would fit with the reduced
room capacities. The Office of Facilities Planning and Management measured over 350 spaces to determine
the reduced capacity of rooms including traditional
classrooms as well as many event spaces. A room that
once fit 150 people could only hold 30, and an early analysis showed that only 60% of classes scheduled could support
the updated room capacity. In other words, 19,000 enrollments would not have a place to sit in the newly sized rooms. SMU either needed more classrooms or a different plan.
We assembled a talented team, and named it the Data Ninjas. The team included staff members from Enrollment Services, Facilities, Institutional Research, and OIT staff including data scientists, database administrators, and programmers. The team sought to build an algorithm and dataset that could move classes to use space most efficiently. Our high-performance computing cluster, ManeFrame II, was unleashed on the problem as the ruleset was constructed and refined.
The team worked many long hours iterating to find the best possible solution while encountering data obstacles. For quite some time, the problem seemed unsolvable. In one early iteration, ManeFrame II ran for

   3   4   5   6   7