The Discrimination You Cannot Avoid

[Ankesh]

Originally Posted by Sandi Bowman

It's no more impossible to evaluate the probabilities of an individual than that of a group...

Thanks Sandi.

If you can - could you please elaborate on the above? How to evaluate probabilities for individuals?

I don't know much about insurance analysis - but it seems interesting to me.

Because I always thought - with my limited knowledge on the subject that - the bigger the group, the better the prediction / probability.

(In a few multivariate tests I've run online for website analysis - that has been the case. Can't predict if the next click will result in a sale or not. But can predict how many sales will I get in the next 100 clicks. And this prediction gets better as more clicks are tracked. I assumed that it would be the same for insurance too. But as I said - my knowledge on the subject is limited and maybe I missed something...)