CPC G06Q 40/12 (2013.12) [G06N 20/00 (2019.01)] | 31 Claims |
1. A method comprising:
(a) extracting, by one or more processors, a set of features from a dataset comprising a plurality of accounts, wherein the set of features are associated with at least one of a plurality of account holders of the plurality of accounts or at least one of a plurality of account variables of the plurality of accounts;
(b) applying, by the one or more processors, a trained algorithm to the set of features to determine (i) a risk score for each of the plurality of account holders and (ii) a feature importance value for each of the set of features associated with the determined risk score;
(c) selecting, by the one or more processors, at least a subset of the plurality of account holders for investigation, based at least in part on risk scores of account holders of the at least the subset; and
(d) outputting, by the one or more processors, (i) the at least the subset of the plurality of account holders selected in (c) and (ii) for each of the at least the subset of the plurality of account holders, at least a subset of the set of features that contribute most to a money laundering risk score of the account holder.
|