Title: Placing Incoming Students in Computer Science, Mathematics, and
Statistics

by Henry M. Walker, Grinnell College

ABSTRACT
Colleges utilize various methods of placing students, but many methods are
time intensive, have limited scope, or lack precision. The placement system
described here resolves many of these issues using a PHP based inference
engine with extensively-researched rules. The system's placements compare
favorably with those created manually by faculty, and students perform well
in the system-recommended courses. Scripts store placements in a MySQL
database and later generate individual LaTeX-based letter for each
student. The scripts from this project run efficiently, follow established
software-engineering principles, and are easily modifiable. The project
automates every step of the process from loading student data into the
database to generating individual letters for students.

In recent years, expert systems have largely been replaced by neural
networks, and my research has explored the possibility of using this
technology to address the placement problem.  Although interesting,
the talk will consider why such an approach, related to Big Data, is
inappropriate for this type of problem and should never be used.

Reference:  The Journal of Computing Sciences in Colleges, Volumn 27,
Number 1, October 2011, pp. 24-31.