How to do distribution learning in AgenaRisk