Professor's Pet: You are just a number

PYGHACK 2018

Introduction

Professor's Pet is a numerical model that will be able to rank classes within an academic department at UIUC, using Prof. Fagen-Ulmschneider's dataset of grade point averages of offered classes.

Our algorithm is a adaption from the chess ELO rating, as we matched classes between each other as well as students against each other. We call our metric the competitiveness rating.

Our Algorithm
Input Data Models:
class student studentclassentry
class_name student_id student_id
elo elo class_name
gpa gpa
Purpose:

A general purpose algorithm that calculates the competitiveness index, a pseudo-elo based score that objectively determines the difficulty of a class or professor or the ability of a student to perform in their respective major.

Results

Using the average GPA of 114 past and present ECE classes, we were able to come up with a competitiveness rating for each of them, and in the graph below show that there is a trend that classes with higher average GPAs tend to have higher ratings:

Among present classes, these are some of the hardest and easiest ECE classes, with higher ratings being easier, and lower ratings being harder. With a cursory glance, we can see that some notorious classes have a low rating due to low average GPA, and advanced graduate seminars are easier:

Class Number and Name Competitiveness Rating Average GPA
ECE 586: Game Theory 1224.629613 3.738095238
ECE 590: Grad Seminar in Special Topics 1213.443122 3.7111233
ECE 544: Topics in Signal Processing 1201.389846 3.609375
...
ECE 391: Computer Systems Engineering 762.7012135 2.459546926
ECE 313: Probability w/ Engineering Applications 674.8997736 2.095202585
ECE 210: Analog Signal Processing 652.7759941 2.088134569

We also took a look at the competitive rating by professor:

Appendix:

Derived update rule

yup steps

Built With

Share this project:
×

Updates