Pros
– Removes bias caused by change in mix of business
– Gives standard error of the estimate, mean vs variance
– Pure effects of independent variables on the dependent variable can be observed
– Assumption that “error terms have a variance independent of the mean” does not need to be made
– Overcomes shortcomings of standard Linear Modelling
Cons
– High powered computing hardware required
– Processing time
– Large number of trial and error runs
– Difficult to review
– Require considerable experience and large portfolios to build in depth models
