Nothing new. Correct application of random and stratified sampling as well as competent machine learning. I started working with that in the early 90s. And, as is the case with all probabilistic modeling, it is very often wrong when the assumptions are wrong. The only thing new is that with today's hardware, Monte Carlo simulation is much faster and the random number generators are more efficient. So, it is possible to get results faster and more accurately than in the past.
And of course big data - storage, collection and analysis is faster and easier.
Last edited by Ezekiel; 07/07/16 02:26 PM.
