Chapter 12 Predictions and Model Goodness

Predictions have a wide range of applications, and in many cases we are not interested in inference, i.e. we are not interested in the parameter values and their interpretation. The python libraries we consider here, statsmodels and sklearn offer easy approaches for predictions, but we start with manual computation, just to make it clear how the models actually work. We spend more time on linear regression, in case of logistic regression we stress more the different types of predictions–probabilities and categories.

Libraries we use in this section: