Stress testing incorporating causal dependency [11]

Go to: Summary | Previous | Next   
Bullet points include: Such a model is usually easier to refine E.g. suppose the sprinkler is replaced by a cleverer one that does not come on if it is already raining This will alter just one element of the causal dependency, so if the model had been structured around the causal dependencies then only a minor tweak would be needed But if the model had been purely associative then we would now need to re-estimate lots of joint probability elements

Contents | Prev | Next | ERM Lecture Series

Desktop view | Switch to Mobile