As machine learning becomes ubiquitous, we are increasingly likely to incorporate additional features from new datasets into existing predictive models. Yet we may have to change how we conceptualize risk pooling in the insurance domain. In the December 2023 issue of “The Actuary Magazine,” published by the Society of Actuaries, Milliman’s Michael Niemerg argues how more data is not always better. The article discusses:
- A simple example of predicting whether it’s currently raining
- A plausible scenario in which a high credit score and routine dental visits predict health outcomes
- Shapley additive explanations values
- How opportunity costs and return on investment matter