Hi there,
It is my understanding that sometimes models are used for the purposes of business selection - for example, you might build a model that says that so-and-so potential insureds that have characteristics xyz are undesirable and hence should no longer be written on the company's books.
However, should such a model be implemented, the company would no longer be collecting data on this population. I was thinking that if the company wanted to write the business at some point in the future, it would no longer have adequate data available for modeling.
Is this a legitimate concern or is there nothing to worry about?
It is my understanding that sometimes models are used for the purposes of business selection - for example, you might build a model that says that so-and-so potential insureds that have characteristics xyz are undesirable and hence should no longer be written on the company's books.
However, should such a model be implemented, the company would no longer be collecting data on this population. I was thinking that if the company wanted to write the business at some point in the future, it would no longer have adequate data available for modeling.
Is this a legitimate concern or is there nothing to worry about?
Feedback between model use and availability of data