You are hired as a data scientist for an insurance company. Every day the company receives a large number of accident claims. Among them, fraud claims are not rare. Examples of fraud may include fake claims for injuries that has never occurred, inflating claim which does not match the actual damage. Not able to detect the fraud will cause enormous loss to your company. You are given the records of all insurance claims, including the details of the claim, whether it is classified as fraud or not, in the past 12 months. One of your job responsibilities is to help detect the fraud.

(a) Name three (3) machine learning approaches/models that you think are appropriate for solving this problem.

(b) The confusion matrix is computed as below, based on one of the approaches you selected.

Transcribed Image Text:

True Fraud Status No Yes Total Predicted Fraud Status No 9,500 30 9530 Yes 250 80 330 Total 9,750 110 9,860 True Fraud Status No Yes Total Predicted Fraud Status No 9,500 30 9530 Yes 250 80 330 Total 9,750 110 9,860

Answer rating: 100% (QA)

a Three machine learning approachesmodels that are commonly used for fraud detection problems are 1 ... View the full answer