Logistic Regression

Locally Weighted Regression Locally weighted regression is a way to modify linear regression and make it fit very non-linear functions, so you aren’t just fitting straight lines. What if the data isn’t just fit well by straight line? how we can addressing out this problem? we wanna use the idea called locally weighted regression or locally weighted linear regression or LWR. So it’s pretty clear what the shape of this data is, but how do you fit a curve that looks like this? It’s actually quite difficult to find features, is it $$ \sqrt x $$ or $$ \log x $$ or $$ x^\frac23 $$ and what is the set of features that lets you do this? ...