Machine learning for fraud preventation across payment types

A system comprising:

• one or more processors, and
• at least one memory storing instructions that, when executed by at least one processor among the one or more processors, cause the at least one processor to perform operations comprising:
• accessing historical transaction data, for a transaction network;
• applying two or more transformations to one or more portions of historical transaction data by at least adding an indicator, to the historical transaction data, that a corresponding transaction was performed from a known bot IP address;
• generating a training data set by (1) mapping a subset of fields of the piece of the historical transaction data to corresponding fields of a plurality of fields in an event data structure based on a first mapping, of a plurality of mappings between a plurality of transaction types and the event data structure, for a first transaction type associated with the piece of the historical transaction data and (2) appending a label, to the event data structure, indicating an outcome for the corresponding transaction represented in the event data structure,
• wherein the plurality of transaction types includes the first transaction type and a second transaction type, and
• wherein the plurality of mappings includes:
• the first mapping, and
• a second mapping, for the second transaction type, that is different from the first mapping;
• and
• retraining a machine learning model using the training data set after the machine learning model is trained previously, after applying the one or more transformations to the piece of the historical transaction data and after generating the training data set by mapping the subset of fields of the piece of the historical transaction data to the corresponding fields of the plurality of fields in the event data structure.

A non-transitory machine-readable medium, storing instructions which, when executed by a machine, cause the machine to perform operations comprising:

• accessing historical transaction data, for a transaction network;
• applying one or more transformations to a piece of historical transaction data by at least adding an indicator, to the historical transaction data, that a corresponding transaction was performed from a known bot IP address;
• generating a training data set by (1) mapping a subset of fields of the piece of the historical transaction data to corresponding fields of a plurality of fields in an event data structure based on a first mapping, of a plurality of mappings maintained between a plurality of transaction types and the event data structure, for a first transaction type associated with the piece of the historical transaction data and (2) appending a label, to the event data structure, indicating an outcome for the corresponding transaction represented in the event data structure,
• wherein the plurality of transaction types includes the first transaction type and a second transaction type, and
• wherein the plurality of mappings includes:
• the first mapping, and
• a second mapping, for the second transaction type, that is different from the first mapping;
• and
• retraining a machine learning model using the training data set after the machine learning model is trained previously, after applying the one or more transformations to the piece of the historical transaction data and after generating the training data set by mapping the subset of fields of the piece of the historical transaction data to the corresponding fields of the plurality of fields in the event data structure.