Much of this book has been involved with classification tasks, where the objective of the analysis is to fit a data point to one of a number of predefined classes or labels. When classifying data, you are able to judge your algorithm's accuracy by comparing predictions to true values; a guessed label is either correct or incorrect. In classification tasks, you can often determine the likelihood or probability that a guessed label fits the data, and you typically choose the label with the maximum likelihood.
Let's compare and contrast classification tasks to regression tasks. Both are similar in that the ultimate goal is to make a prediction, informed by prior knowledge or data. Both are similar in that we want to create some kind of function or logic that maps input values to output values, and make that mapping function both as accurate...