Andrew on CSE Award Criteria

I can't make the TC meeting so here's roughly what I said last time this was discussed at the TC

Any comparison of performance between program is going to be inexact. The criteria used for the awards are the fairest we could discover.

When these awards were introduced we tested criteria by looking at the awards that would have been given over the 5 previous years.

When generating each year's award lists I also generate lists with looser criteria to look for strong candidates omitted because of some defect in the criteria.

CE, SE & CS students do roughly the same number of COMP courses each year.

Comparing the best subset of COMP courses reduces the effects of differences between the COMP courses taken (the subsets are for years 1,2,3,4 respectively 1711/1721, 18 uoc, 24uoc, thesis+12uoc).

Including SENG courses in the comparison would assist SE students. This would be mostly unfair in my view as their best subset would now be chosen from a larger set than CE/SE students. It would also increase heterogeneity of the candidate courses.

BINF&SENG courses are included if the students hasn't done sufficient COMP courses to fill the subset. This is important for 3647/2 students but might affect a few others with unusual enrollments.

3647 is more problematic So far they seem to have been treated OK but there is only a very small number of students involved.

These criteria have the virtue of relatively simplicity. My experience is that adding complexitity just makes problems less obvious.


School of Computer Science & Engineering
The University of New South Wales
Sydney 2052, AUSTRALIA
Phone: +61 2 9385 6876

UNSW's CRICOS Provider No. is 00098G
Last updated: