We will determine final course grades as follows.
- Drop, say, 25% lowest homework problems grades. Every
homework would have 3 problems, each worth 100 points. So,
every student would have roughly 24-30 homework problem
submissions (assuming they submit all of them). Of them, the
lowest t would be dropped.
The value t is computed as follows: It is the maximum of 6 and
25% of the homework problems that would be
graded. Thus, if 27 homework problems would be graded, then
t=rounded_up(27/4) = 7.
- Compute everyone's adjusted course average (ACA).
Course work is weighted as follows:
- 20% for homework,
- 50%: midterms: 2 midterms * 25% for each midterm.
- 30% for the final exam.
- In rare cases a few students performing exceptionally
better than all the other students in a class *might* get an
A+. Since A+ is equal to an A as far as GPA in the University of
Illinois, this has no impact on GPA. The only guaranteed way for
getting an A+ is to get a perfect score on all exams, and homeworks.
- Grads and undergrads are graded on the same scale.
- Getting an F:
- Failing to attend 70% of the lectures.
- Anyone with a homework average below 40% or a
ACA below 30%, automatically gets an F. (You can fail in
other ways too, see below.)
- Any student with an exam average <= 25%.
This are not the only ways to fail this class.
- Compute basic statistics of the distribution of the ACA. This
would be NOT be done separately for grad and undergrad
students. Everybody would be graded using the same grading curve.
- Determine letter grade cutoffs, excluding extreme outliers
at both ends of the curve.
- The median is a borderline B/B+, and each standard deviation
is worth 2/3 of a letter grade. Thus, the B+/A- cutoff is 1/2 a
standard deviations above the median. That is, the range that
corresponds to the grade B itself has length of 1/3 of a standard
deviation. Thus:
B- = [Median - SD, Median- SD/2)
B = [Median - SD/2, Median)
B+ = [Median, Median + SD/2)
...and so on.
(The final grading would be at least as generous as this.)
- The above is just a guideline - the thresholds would be
manipulated as the instructor see fit (within reason).
- Compute final letter grades from ACAs, except for the
outliers from previous steps.
- Adjust grades (only upwards!) as the instructor sees fit.
There is almost no extra credit in this course. There would be
questions where we expect very few students to answer correctly, and
one can think about them as extra credit.
A few students that participate extensively on piazza and answer
many questions correctly might get a grade bump in the end of the
semester. Naturally, useless participation (i.e., providing
incorrect answers, asking not so useful questions, making useless
remarks, etc) would not be rewarded.
Note: We are grading undergrads and graduate students on the
same curve
Last modified: Tue 2021-10-12 19:41:00 UTC 2021 by Sariel Har-Peled