Low latency bank card type prediction system for estimation of interchange codes during transaction processing

A non-transitory computer-readable storage medium storing instructions that, when executed by a computer, cause the computer to perform a method for using a computer system to predict interchange charges corresponding to credit card transactions, the method comprising:

• via a server:
• retrieving a historical completed transactions set from a database that corresponds to a plurality of merchants, where each completed transaction in the set comprises features, a bank identification number (BIN), and a corresponding true interchange code;
• extracting the features from the transaction data;
• transforming all BINs in the set into a corresponding plurality of BIN features that comprise probabilities of a given BIN falling into each of the corresponding plurality of BIN features, where each of the probabilities are determined by analyzing the historical completed transactions set;
• creating a first training set comprising all features, all pluralities of BIN features, and all true interchange codes associated with the historical completed transactions set;
• training a random forest model using the first training set, wherein the training comprises rounding of all BIN features and non-discrete ones of the features, and generating a second training set comprising rounded BIN features, rounded features, discrete ones of the features, and the true interchange codes;
• training the random forest model using the second training set to generate a trained random forest model for prediction of the interchange codes;
• receiving transaction data corresponding to a transaction from a merchant point-of-sale terminal;
• extracting transaction features from the transaction data, wherein the transaction features comprise a transaction type;
• translating and rounding non-discrete values of the transaction features into rounded transaction features according to cut points determined by training the random forest model using the first training set;
• executing the trained random forest model using the rounded transaction features as inputs to generate a predicted interchange charge for the transaction, wherein the predicted interchange charge comprises a function of the interchange codes;
• computing a holdback amount for the transaction based upon the predicted interchange charge;
• transmitting the transaction data and holdback amount to a payment processor for settlement of the transaction;
• and
• responsive to periodic updates to the interchange codes by one or more issuing banks, employing subsequent transactions processed according to updated interchange codes to retrain the trained random forest model to predict updated interchange charges, to compute updated holdback amounts, and to transmit the updated holdback amounts to the payment processor for settlement of new transactions.

A system for predicting interchange charges corresponding to credit card transactions, the system comprising:

• a server, configured to:
• retrieve a historical completed transactions set from a database that corresponds to a plurality of merchants, where each completed transaction in the set comprises features, a bank identification number (BIN), and a corresponding true interchange code;
• extract the features from the transaction data;
• transform all BINs in the set into a corresponding plurality of BIN features that comprise probabilities of a given BIN falling into each of the corresponding plurality of BIN features, where each of the probabilities are determined by analyzing the historical completed transactions set;
• create a first training set comprising all features, all pluralities of BIN features, and all true interchange codes associated with the historical completed transactions set;
• train a random forest model using the first training set, wherein the training comprises rounding of all BIN features and non-discrete ones of the features, and generating a second training set comprising rounded BIN features, rounded features, discrete ones of the features, and the true interchange codes;
• train the random forest model using the second training set to generate a trained random forest model for prediction of the interchange codes;
• receive transaction data corresponding to a transaction from a merchant point-of-sale terminal;
• extract transaction features from the transaction data, the transaction features comprising a type of transaction;
• translate and round non-discrete values of the transaction features into rounded transaction features according to cut points determined by training the random forest model using the first training set;
• execute the trained random forest model using the rounded transaction features as inputs to generate a predicted interchange charge for the transaction, wherein the predicted interchange charge comprises a function of the interchange codes;
• compute a holdback amount for the transaction based upon the predicted interchange charge;
• transmit the transaction data and holdback amount to a payment processor for settlement of the transaction;
• and
• responsive to periodic updates to the interchange codes by one or more issuing banks, employ subsequent transactions processed according to updated interchange codes to retrain the trained random forest model to predict updated interchange charges, to compute updated holdback amounts, and to transmit the updated holdback amounts to the payment processor for settlement of new transactions.