Q.5. Income Tax file records are to be examined to develop a model for predicting fraudulent ones. 2% of the records in the historical database were judged as fraudulent. Undersampling was used for the majority class to provide a balanced sample to develop a predictive model to predict frauds and when applied to this sample (N=1000), the model correctly classifies 430 frauds and misclassifies 70 actual frauds. It also correctly classifies 320 non-frauds and wrongly classifies 180 actual non-frauds.

(i) Find the adjusted misclassification rate for undersampling.

(ii) What percent of new records would you expect to be classified as non-frauds?

[3.5+3.5 =7 Marks]