5.9 Cross-Validation on Classification Problems

• Previous examples have focused on measuring cross-validated test error in the regression setting where \(Y\) is quantitative.
• We can also use cross validation for classification problems (where \(Y\) is qualitative). Here we use the number of misclassified observations (instead of the MSE) to quantify test error.
• The LOOCV error rate takes the form:

\[CV_{n} = \frac{1}{n}{\sum_{i=1}^{n}}Err_{i}\] where \(Err_{i} = I(Y_{i}\neq\hat{Y}_{i})\)