5 comments

[ 4.3 ms ] story [ 35.7 ms ] thread
Still need that sigmoid assumption, and that's a lot to swallow.
Yes, Logistic regression applies when you're modeling a binary random variable and expect the log-odds to be linearly related to your regressors.
A very nice explanation. It's reminds me of Andrew Ng's lectures. Very impressive that the author seems to be a high school student - I'm only learning about these concepts now in Masters.
indeed, it blew me away that the author is a high school student.