@osmelbrito - To get a definitive answer I think you will have to talk to SAS Tech Support as this is a very specialist topic.
I'm not a user of SAS AML, but I suspect an MLS would be constructed in a similar way as a credit application score or account behaviour score. That is you look at a whole bunch of measures related to the customer's personal, bank account and transaction details. For example a pattern of large cash transactions would be a higher risk than electronic transactions with known sources. You would then decide on those that were most predictive based on real AML transactions. This would be turned into a scorecard where the higher the score, the higher the probability of being money laundering. An MLS would need to be constructed using your client's data and experience of AML - I guess this would be a bank or other financial institution.
