Question

A data mining routine has been applied to a transaction dataset and has classified 88 records as fraudulent (30 correctly so) and 952 as non-fraudulent (920 correctly so). Construct the classification matrix and calculate the error rate, sensitivity, and specificity.

Answer 1

1. Classification Matrix:

Actual Class 1 0

1 30 58

0 32 920

2.Overall error rate (err) 8.65%

Accuracy% 91.35%

Step-by-step explanation:

To Construct the classification matrix and to calculate the error rate

1.CONSTRUCTION OF THE CLASSIFICATION MATRIX

Classification Matrix:

Classification Confusion Matrix

Predicted Class

Actual Class 1 0

1 30 58

(88-30=58)

0 32 920

(952-920)

Calculation for the error rate

Using this formula

Overall error rate (err) = Sum of Misclassification transactions/Total transaction

Let plug in the formula

Overall error rate (err) =[(88-30)+(952-920)]/(952+88)

Overall error rate (err) = (58+32)/1040

Overall error rate (err) =0.08653*100

Overall error rate (err) = 8.65 %

Accuracy% = 1-err

Accuracy%=100%-8.65%

Accuracy%= 91.35%