Why Classification

As Gendler states (2011), “classifying objects into groups allows us to proceed effectively in an environment teeming with overwhelming detail.”

Random Baseline

In order to compute the accuracy of a random baseline for a multi-class classfication you can simply: \[acc = P(y=0) \times P(\hat{y}=0) + P(class is 1) \times P(you guess 1) \] where P(y=1) means the probability that the class is actually 1, and \(P(\hat{y}=1)\) means tne probability that you guess 1.

Majority-Class Classifier

This a baseline which is equal or better than random baseline (False. It can be worst) in which we always predict the majority class. It’s also called zero rule (ZeroR or 0R).