Saturday, October 27, 2012

A Tableau for UCSD Course Grade Distributions

I come across this information completely by accident. I am quite surprised (and proud) that the school makes its class grade distributions data public (link here). A quick googling on grade distribution shows that other schools – Indiana University for example (link here) – also release this type of data.

I don’t think the school user interface is friendly to data exploration, so I wrote a short Perl script to pull and parse the data and dump it into a tableau. The presentation here is quite basic, but it should still be useful for students. If I have some time later, I’ll look into the data more and see if I find anything interesting. For example, over the sample period (since 2007), Math classes on average has the second lowest class GPA (2.752, 705 observations) among all subjects. SDCC (I do not know what this stands for, but appears to relate to writing) has the lowest average class GPA (2.213, 128 observations). Math on average also gives F to 5.139% of its classes (second highest). Of course, none of what I have found is all that surprising. Math is hard. Writing is hard. That’s why students are tested these subjects on the SAT.

If you are a UCSD student that happens upon this blog entry – just remember that correlation does not equal causation! For example, you will find from the data below that Econ100BH, an honors class, has a class GPA average of 3.907. That is definitely *not* because the class is easy. More likely, the students who tend to get A self select into taking that class.

If you have never seen a Tableau, here’s the company website (link). Basically, if you hover over things below, an information box will show up. When you click on things – for example, the professor’s name – the Tableau shows all the courses he or she has taught since 2007 and what the class GPA is for each of those classes. I set the default to ECON (Economics), but you can select different subject codes (lower right). 

No comments:

Post a Comment