I am trying to predict how hard it will be to get payment for final arrears from a customer. The hardness is measured in a scale of 0 - 5 (0 meaning easy, no credit collection action needs to be taken, customer just pays. 5 is hardest, and means a significant number of credit collection actions had to be taken before the customer paid or possibly the customer didn't pay at all ever.)
I have approximately 8 fields for input data. Fields include
- customer credit rating information (-200 (bad) to +200 (good))
- was it ever necessary to take legal action against the customer (Y/N)
- average number of phone calls per calendar month to customer (more is worse as implies more involvement necessary to get customer to pay bills)
Which Oracle analytics data mining algorithm would I throw the training data at to see how well the effort required to collect a customer's final debt can be predicted?
If you are casting this a classification problem, then you can try all of the Classification algorithms contained in the Classification build node. It includes a internal set of test results to show you which algorithm performed the best.