There are currently two modes for an assignment collection: pass all or pass one. I would like to generalize this to pass n out of m assignments (where n <= m). n == m corresponds to ‘pass all’, n == 1 corresponds to ‘pass one’. n == (m - 1) corresponds to a collection of which one can be skipped (or alternatively the last assignment compensates for another assignment missed earlier).

The grading scheme should be to select the n highest grades from the submissions and computer their (weighted) average. If fewer than n assignments have been submitted the grade should be maxed at the passing grade minus a half grade point.

Submitted by Eelco Visser on 29 December 2014 at 23:55

Elmer: I needed this last year. I really need it now.

Subassignments with weight 0.0 are to be ignored for both n and m.

Edit: Just decided to take all sub-assignments in consideration for n. In case the weight of one sub-assignment is set to 0.0, it won’t contribute to the grade, but it contributes to the n passed assignments.

The grades that contribute to the avg grade of the collections are the n highest grades that have pass. If there are not enough passed assignments, it also takes the remaining failed submissions in order of highest grade first to calculate the average avg.The eventual grade will be:
if( passed n assignments) -> avg
else -> min( avg , minimum-0.5)

(1) hidden assignments should be ignored

(2) I don’t think it makes sense to include assignments with weight 0

(3) otherwise agreed

Not sure if this belongs here, but support for O/V assignments, which need to be passed for the course, but do not contribute to the overall grade, do not fit very well with the current handling of assignments with 0.0 weights but passing requirements.