Well they do say they calculate propensity scores, but the whole language of the blog post is very hand-wavy. Warning over and over again that something is not causal proof, and then dramaticizing it as if it were proof is just unprofessional.
If you have a fraud model, just show the model and the data and the validation - everything else is marketing fluff.
Comments
I'm kind of irritated that the article doesn't say anything about probabilities.
Given they have enough data, at some point it's perfectly reasonable to have cars with 5+ crashes per 12 months - just because of chance.
This is exactly why statistics was invented, damnit!
I think “in the exact same way with all the known indicators of insurance fraud” adds a couple orders of magnitude to the expected by-chance odds.
Well they do say they calculate propensity scores, but the whole language of the blog post is very hand-wavy. Warning over and over again that something is not causal proof, and then dramaticizing it as if it were proof is just unprofessional.
If you have a fraud model, just show the model and the data and the validation - everything else is marketing fluff.