I took this class last year (some variations, less material covered more deeply) and it was a great survey course for someone who had programming experience but was lacking in stats and ML background!
I’m a huge fan of the Extension program. It enables access to great courses and if you do well and are in a degree program you can get special student status opening up the college and grad school class listings.
I disagree with others here --- I wasn't a huge fan of the class.
If you took the class, you definitely had to get used to coding in python, and getting used to manipulating data in python.
However, when it came time to do anything more interesting than plotting data, the course was very shallow --- many homework assignments boiled down to "fill in this space with a function that takes this as input and outputs that". While, again, that would totally be useful to someone learning basic scientific python, you don't get a good understanding of the statistics.
As someone who already had a background in python, I didn't feel I learned as much.
Also, the course walks over a few common ML algos and provides some predefined datasets as well.
Fairly good course, in fact, developers in my team build custom models for our internal hackday only by following this course and a few open datasets. I guess the datasets were from here:
https://dataturks.com/projects/trending
Similarly structured course on data visualization w/ d3. covers the fundamentals of the art (mostly Tufte) alongside a pretty decent intro to using d3.
19 comments
[ 0.19 ms ] story [ 53.1 ms ] threadI was a TF and gave a couple of lectures. AMA.
What's your opinion on it?
How does it compare to 109?
If you took the class, you definitely had to get used to coding in python, and getting used to manipulating data in python.
However, when it came time to do anything more interesting than plotting data, the course was very shallow --- many homework assignments boiled down to "fill in this space with a function that takes this as input and outputs that". While, again, that would totally be useful to someone learning basic scientific python, you don't get a good understanding of the statistics.
As someone who already had a background in python, I didn't feel I learned as much.
Fairly good course, in fact, developers in my team build custom models for our internal hackday only by following this course and a few open datasets. I guess the datasets were from here: https://dataturks.com/projects/trending
http://www.cs171.org/2017/index.html