Separating the training set into two sets of policies with and without claims
NClaim <- subset(DataTrain, DataTrain$claim_nb == 0)
Claim <- subset(DataTrain, DataTrain$claim_nb > 0)