Banks are interested on developing models that are able to stop fraudulent transactions. Lets say you are hired in a maj
Posted: Thu Jun 02, 2022 7:49 am
Banks are interested on developing models that are able to stop
fraudulent transactions. Lets say you are hired in a major
American bank, where you will develop a system that helps to
stop frauds in a credit card portfolio.
Suppose that in your first task you create a data set containing
1 000 covariates and where the number of frauds is only 0.001 of
the total sample size, which is 500 000. Discuss why the true
negative rate is not an appropriate metric in this case.
fraudulent transactions. Lets say you are hired in a major
American bank, where you will develop a system that helps to
stop frauds in a credit card portfolio.
Suppose that in your first task you create a data set containing
1 000 covariates and where the number of frauds is only 0.001 of
the total sample size, which is 500 000. Discuss why the true
negative rate is not an appropriate metric in this case.