-
True Positives: The number of points that the classifier correctly predicts as positive.
$$TP=n_{11}=|{x_i|\hat{y_i}=y_i=c_1}|$$ -
False Positives: The number of points the classifier predicts to be positive, which in fact belong to the negative class.
$$FP=n_{21}=|{x_i|\hat{y_i}=c_1 \land y_i=c_2}|$$ -
False Negatives: The number of points the classifier predicts to be in the negative, which in fact belong to the positive class.
$$FN=n_{12}=|{x_i|\hat{y_i}=c_2 \land y_i=c_1}|$$ -
True Negatives: The number of points that the classifier correctly predicts as negative.
$$TN=n_{22}=|{x_i|\hat{y_i}=y_i=c_2}|$$
True 是预测正确的，False 是预测错误的，True 和 False 后面是预测的结果。
-
Error Rate
$$Error\ Rate=\frac{FP+FN}{n}=1-Accuracy$$
对于每个分类，可以将当前分类作为 Positive，所有其它分类作为 Negative。TP 为正确归类到当前分类，TN 为正确归类到其它分类，FP 为错误归类到当前分类，FN 为错误归类到其它分类。