Binary Classifications

True Positive

It is an instance where the model correctly predicts the positive class i.e, the Model predicts a true value given that the answer is actually true.

• True Positive Rate It is the ratio of the number of True Positive classifications to the actual number of positives.

$$TPR=\frac{TP}{TP+FN}$$

> Note: Also called *Hit Rate* or *Recall*

False Positive

It is an instance where the classifier incorrectly predicts a positive class. i.e, the Model predicts a true value given that the actual answer is false.

• False Positive Rate It is the ratio of the number of False Positive classifications to actual number of negatives.

$$FPR=\frac{FP}{FP+TN}$$

> Note: Also called *False Acceptance Rate (FAR)*

True Negative

It is an instance where the model correctly predicts a negative class.

• True Negative Rate It is the ratio of the number of Negative classifications to the actual number of negatives.

$$TNR=\frac{TN}{TN+FP}=1-FPR$$

> Note: Also called *Specificity*

False Negative

It is an instance where the Model incorrectly predicts a Negative class.

• False Negative Rate It is the ratio of the number of False Negative classifications to actual number of false negatives.

$$FNR=\frac{FN}{FN+TP}$$

> Note: Also called *False Rejection Rate* or *Miss Rate*

Object	ML Term
True Positive	Hit rate / Recall
False Positive	False Acceptance Rate
True Negative	Specificity
False Negative	False Rejection Rate / Miss Rate