Confusion Matrix

The plane below shows points whose marker is the actual class (+ or −); the thick line is a decision boundary: everything to its left is predicted +, everything to its right is predicted −. Drag anywhere on the plane to move the boundary. Each point's color is its outcome, matching the matrix cells.

Data & decision boundary

Confusion matrix (live counts)

TP = True Positive · FN = False Negative
FP = False Positive · TN = True Negative
Scenarios from the notes
\( \text{Accuracy} = \dfrac{\#\text{ right}}{\#\text{ total}} \)  = 
\( \text{Misclassification Error} = \dfrac{\#\text{ wrong}}{\#\text{ total}} \)  = 
\( \definecolor{tp}{RGB}{46,125,50}\definecolor{fn}{RGB}{230,159,0} \text{Recall} = \dfrac{\textcolor{tp}{\text{TP}}}{\textcolor{tp}{\text{TP}}+\textcolor{fn}{\text{FN}}} \)  =    \( \longleftrightarrow \)   \( \definecolor{tn}{RGB}{48,112,183}\definecolor{fp}{RGB}{192,57,43} \text{Specificity} = \dfrac{\textcolor{tn}{\text{TN}}}{\textcolor{tn}{\text{TN}}+\textcolor{fp}{\text{FP}}} \)  = 
\( \definecolor{tp}{RGB}{46,125,50}\definecolor{fp}{RGB}{192,57,43} \text{Precision} = \dfrac{\textcolor{tp}{\text{TP}}}{\textcolor{tp}{\text{TP}}+\textcolor{fp}{\text{FP}}} \)  = 
• Recall: high when FN is low compared to TP   • Precision: high when FP is low compared to TP