Grading and Peer evaluation

How we do grading

I'd like to make two central points below

1) The peer grading scores your work received does not influence your grade

I strongly believe that you should be solving complex problems. Problems that have many correct solutions, just like in the real world. This means problems that you can't pose as multiple choice assignments, you can't get a computer to grade this kind of work (yet - maybe with deep learning, but that's another story).

This class, however, has around 100 students. So how do we give you good feedback? The answer is peer evaluations. By crowd-sourcing evaluations, you guys can get feedback on your work quickly and when you average over many other students, the quality is as high as you'd get from TA's. There are lots of pro/cons related to peer evaluation and I discuss those elsewhere on these pages.

But an important thing to realize is that how your peers evaluated you, does not determine your grade! (We do look at the quality of your peer evaluations of others' work. And the quality of your evaluations of other people's assignments, is reflected in the grade, see below).

2) So how does the grade come about?

When it's time to do the grading, the TAs and I get together. And we set it up so that at least two of us look at each assignment (including the final project assignments). Then we discuss each assignment as a group and write down a numerical assessment. We also take into account the quality of your peer evaluations.

Each grade is then based on those numerical assessments. The grade is a holistic evaluation of your work in the entire course, but as a rule of thumb 50% of the grade is due to Assignment 1 and 2, while the other 50% are due to your final project assignments and peer assessments.