Logistic regression -> Binary evaluation Features
Transform linear regression to probability, meaning output between 0 and 1, using sigmoid function
Cost function, calculate error
Gradient to understand the parameters that minimize the error Having a lot of local minimums is bad