12.5 Vocabulary terms

Also other names for the same term, and formulas.

• True positive ($n_{11}$)

• True negative ($n_{22}$)

• False positive, Type I error ($n_{12}$)

• False negative, Type II error ($n_{21}$)

• True positive rate (TPR), Recall, Sensitivity, probability of detection, power. P(predicted positive | total positive) $\frac{\# True Positive}{\#Condition Positive}$

• True negative rate (TNR), Specificity, selectivity. P(predicted negative | total negative) $\frac{\# True Negative}{\# Condition Negative}$

• False positive rate (FPR), fall-out, probability of false alarm. $\frac{\# False Positive}{\# Condition Negative}$

• False negative rate (FNR), miss rate. $\frac{\# False Negative}{\# Condition Positive}$

• Prevalence. $\frac{\# Condition Positive}{\# Total Population}$

• Accuracy. $\frac{\# True Positive + \# True Negative}{\# Total Population}$

• Balanced Accuracy: $[(n_{11}/n_{.1}) + (n_{22}/n_{.2})]/2$ - Adjusts for class size imbalances

• Positive Predictive Value (PPV), Precision. P(true positive | predicted positive) $\frac{\# True Positive}{\# Predicted Condition Positive}$

• False discovery rate (FDR). $\frac{\# False Positive}{\# Predicted Condition Positive}$

• False omission rate (FOR). $\frac{\# False Negative}{\# Predicted Condition Negative}$

• Negative predictive value (NPV). $\frac{\# True Negative}{\# Predicted Condition Negative}$

• F1 score. The harmonic mean of precision and recall. This ranges from 0 (bad) to 1 (good): $\frac{2 * (Precision * Recall)}{Precision + Recall}$