Mathematics/Computer Science Placement
Problem:
Place incoming students, based on HS transcript, standardized scores (SAT, ACT, AP, IB)
Before 1994, done by faculty: time consuming, based mostly on intuition, experience
Results sent only to advisor with no comments regarding basis
Experience showed faculty at least as accurate as MAA placement test
Scholarship
Approach: Automate process using basic research on expert systems from AI
Inference engire based on
TMYCIN
from AI Lab at UT-Austin
Teaching
Independent project with 2 students (one now Director of Product Management at eBay)
Used as example of expert systems in several courses
Service
Now full letter sent to both student and advisor, including basis for placement
Several follow-up studies to compare placement and subsequent student performance
Status
Current system generally better than faculty, prior to student arrival on campus
Paper co-authored with students appeared in 1994
Link to Prospectives' Page from Department
created 23 April 2005
last revised 29 April 2005
previous
next
