Confusion Matrix

Dog 23 12 7
Cat 11 29 13
Rabbit 4 10 24
Actual Predicted
Dog Cat Rabbit
Dog 23 12 7
Cat 11 29 13
Rabbit 4 10 24
Multiclass Confusion Matrix

Calculate the value of Accuracy Rate, Sensitivity and Specificity.

TPTNFPFN
Dog 23 29 + 13 + 10 + 24 = 76 11 + 4 = 15 12 + 7 = 19
Cat 29 23 + 7 + 4 + 24 = 58 12 + 10 = 22 11 + 13 = 24
Rabbit 24 23 + 12 + 11 + 29 = 75 7 + 13 = 20 4 + 10 = 14
Per-Class Binary Confusion Metrics Table

Accuracy for Dog

Accuracy = (TP + TN) / (TP + TN + FP + FN)

= (23 + 76) / (23 + 76 + 15 + 19)

= 99 / 133 = 0.744 = 74.4%

Sensitivity for Cat

Sensitivity = TP / (TP + FN)

= 29 / (29 + 24) = 0.54 = 54%

Specificity for Rabbit

Specificity = TN / (TN + FP)

= 75 / (75 + 20) = 0.78 = 78%

Confusion Matrix Example-2:

Consider the following 3-class confusion matrix. Calculate precision and recall per class.
Also calculate weighted average precision and recall for classifier.

15 2 3
7 15 8
2 3 45
ABCTotal
A 152320
B 715830
C 234550
Total 242056100

Precision = Correctly Predicted / Total Predicted

Precision of Class A = 15 / 24 = 0.625

Precision of Class B = 15 / 20 = 0.75

Precision of Class C = 45 / 56 = 0.80

Recall = Correctly Classified / Actual

Recall of Class A = 15 / 20 = 0.75

Recall of Class B = 15 / 30 = 0.50

Recall of Class C = 45 / 50 = 0.90

Accuracy = Total Correctly Classified / Total Actual

Accuracy of Classifier = (15 + 15 + 45) / 100 = 0.75

Actual Class A instances = 20 / 100 = 0.2

Actual Class B instances = 30 / 100 = 0.3

Actual Class C instances = 50 / 100 = 0.5

Weighted average precision =

(0.2 × 0.625) + (0.3 × 0.75) + (0.5 × 0.80) = 0.75

Weighted average recall =

(0.2 × 0.75) + (0.3 × 0.5) + (0.5 × 0.90) = 0.75

Solution Done