Key concepts of logistic regression
If you are familiar with linear regression, or read Chapter 7, Linear Regression Models, of this book, you have probably anticipated some of the issues we will discuss in this chapter – regularization, linearity among regressors, and normally distributed residuals. If you have built supervised machine learning models in the past or worked through the last few chapters of this book, then you have also likely anticipated that we will spend some time discussing the bias-variance tradeoff and how that influences our choice of model.
I remember being introduced to logistic regression 35 years ago in a college course. It is often presented in undergraduate texts almost as a special case of linear regression; that is, linear regression with a binary dependent variable coupled with some transformation to keep predictions between 0 and 1.
It does share many similarities with linear regression of a numeric target variable. Logistic regression...
