February 2018
Intermediate to advanced
378 pages
10h 14m
English
Linear regression can be trivially adapted for binary classification: just predict a positive class for all regression outputs above some threshold and a negative class for everything below it. For example, in the following diagram, the threshold is 0.5. Everything with x < 0.5 gets classified as a negative class and everything with x > 0.5 as positive. The line that separates feature values of one class from another is called a decision boundary. With more than one feature, the decision boundary will not be a line but a hyperplane:

Read now
Unlock full access