System, Method, and Computer Program Product for Determining Fraud Rules

A computer-implemented method for generating a set of fraud detection rules based on historical transaction data including both non-fraudulent transactions and fraudulent transactions, the computer-implemented method comprising:

• receiving, with at least one processor, historical transaction data including a plurality of transactions associated with an issuer institution, the plurality of transactions including a plurality of non-fraudulent transactions and a plurality of fraudulent transactions;
• associating, with at least one processor, a tag with each of the transactions, the tag associated with a predetermined transaction type;
• generating, with at least one processor, at least one decision tree having a root node and at least one child node operably connected to the root node;
• determining, with at least one processor, at least one primary rule and at least one set of secondary rules associated with the at least one primary rule based on relationships between correlated features associated with the plurality of non-fraudulent transactions and the plurality of fraudulent transactions;
• assigning, with at least one processor, the at least one primary rule and the at least one set of secondary rules associated with the at least one primary rule to the at least one decision tree to populate the at least one decision tree;
• extracting, with at least one processor, a plurality of rule sets from the at least one decision tree, each of the plurality of rule sets including an ordered set including the at least one primary rule and one or more secondary rules from among the at least one set of secondary rules associated with the at least one primary rule;
• determining, with at least one processor, an ordering of the plurality of rule sets, the ordering based on memory-based efficiency characteristics and fraud-detection-based characteristics associated with each of the rule sets of the plurality of rule sets;
• and
• determining, with at least one processor, a subset of rule sets from the ordered plurality of rule sets against which subsequently received transactions are compared against to determine if the subsequently received transactions are non-fraudulent transactions or fraudulent transactions.

A system for generating a set of fraud detection rules based on historical transaction data comprising both non-fraudulent transactions and fraudulent transactions comprising:

• at least one processor programmed or configured to:
• receive historical transaction data including a plurality of transactions associated with an issuer institution, the plurality of transactions including a plurality of non-fraudulent transactions and a plurality of fraudulent transactions;
• associate a tag with each of the transactions, the tag associated with a predetermined transaction type;
• generate at least one decision tree having a root node and at least one child node operably connected to the root node;
• determine at least one primary rule and at least one set of secondary rules associated with the at least one primary rule based on relationships between correlated features associated with the plurality of non-fraudulent transactions and the plurality of fraudulent transactions;
• assign the at least one primary rule and the at least one set of secondary rules associated with the at least one primary rule to the at least one decision tree to populate the at least one decision tree;
• extract a plurality of rule sets from the at least one decision tree, each of the plurality of rule sets including an ordered set including the at least one primary rule and one or more secondary rules from among the at least one set of secondary rules associated with the at least one primary rule;
• determine an ordering of the plurality of rule sets, the ordering based on memory-based efficiency characteristics and fraud-detection-based characteristics associated with each of the rule sets of the plurality of rule sets;
• and
• determine a subset of rule sets from the ordered plurality of rule sets against which subsequently received transactions are compared against to determine if the subsequently received transactions are non-fraudulent transactions or fraudulent transactions.

A computer program product for generating a set of fraud detection rules based on historical transaction data including both non-fraudulent transactions and fraudulent transactions, comprising at least one non-transitory computer readable medium including one or more instructions that, when executed by at least one processor, cause the at least one processor to:

• receive historical transaction data including a plurality of transactions associated with an issuer institution, the plurality of transactions including a plurality of non-fraudulent transactions and a plurality of fraudulent transactions;
• associate a tag with each of the transactions, the tag associated with a predetermined transaction type;
• generate at least one decision tree having a root node and at least one child node operably connected to the root node;
• determine at least one primary rule and at least one set of secondary rules associated with the at least one primary rule based on relationships between correlated features associated with the plurality of non-fraudulent transactions and the plurality of fraudulent transactions;
• assign the at least one primary rule and the at least one set of secondary rules associated with the at least one primary rule to the at least one decision tree to populate the at least one decision tree;
• extract a plurality of rule sets from the at least one decision tree, each of the plurality of rule sets including an ordered set including the at least one primary rule and one or more secondary rules from among the at least one set of secondary rules associated with the at least one primary rule;
• determine an ordering of the plurality of rule sets, the ordering based on memory-based efficiency characteristics and fraud-detection-based characteristics associated with each of the rule sets of the plurality of rule sets;
• and
• determine a subset of rule sets from the ordered plurality of rule sets against which subsequently received transactions are compared against to determine if the subsequently received transactions are non-fraudulent transactions or fraudulent transactions.