GENERATING A FRAUD PREDICTION UTILIZING A FRAUD-PREDICTION MACHINE-LEARNING MODEL
This disclosure describes an intelligent fraud detections system that, as part of an inter-network facilitation system, can intelligently generate fraud predictions for digital claims to improve accuracy and efficiency of network-transaction security systems. For example, the disclosed systems can utilize a fraud detection machine-learning model to generate a fraud prediction for a digital claim disputing a digital transaction. Indeed, the disclosed systems can identify features of a digital claim and, based on those features, utilize a fraud detection machine-learning model to generate a fraud prediction for the digital claim. Additionally, based on the fraud prediction, the disclosed systems can perform authorizing, remedial, or other actions with regard to the digital claim.
1 . A non-transitory computer-readable medium comprising instructions that, when executed by at least one processor, cause a computing device to:
receive a digital claim disputing a network transaction;
identify one or more features associated with the digital claim;
generate, utilizing a fraud detection machine-learning model, a fraud prediction for the digital claim based on the one or more features; and
provide, for display in a graphical user interface, a visual indicator of the fraud prediction for the digital claim.
2 . The non-transitory computer-readable medium as recited in claim 1 , further storing instructions thereon that, when executed by the at least one processor, cause the computing device to identify the one or more features associated with the digital claim by determining one or more of: an account feature, an automated-clearing-house feature, a computing device feature, a demand draft feature, a transaction feature, a peer-to-peer-payment feature, an identity verification feature, a shared device feature, a shared-internet-protocol-address feature, a customer-service-contact feature, a failed login feature, a password reset feature, a personal identifiable-information-change feature, a linked-claim-dispute feature, a dispute history feature, or a merchant feature.
3 . The non-transitory computer-readable medium as recited in claim 1 , further storing instructions thereon that, when executed by the at least one processor, cause the computing device to identify the one or more features associated with the digital claim by determining one or more of a zip code feature, a merchant category code feature, an account dormancy time feature, a sign in feature, or a transaction-number feature.
4 . The non-transitory computer-readable medium as recited in claim 1 , further storing instructions thereon that, when executed by the at least one processor, cause the computing device to:
determine that the fraud prediction satisfies a moderate-risk fraud prediction threshold; and
based on determining that the fraud prediction satisfies the moderate-risk fraud prediction threshold, provide, for display within in the graphical user interface, the visual indicator by providing a moderate-risk visual indicator of fraud for the digital claim.
5 . The non-transitory computer-readable medium as recited in claim 1 , further storing instructions thereon that, when executed by the at least one processor, cause the computing device to:
determine that the fraud prediction satisfies a high-risk fraud prediction threshold; and
based on the fraud prediction satisfying the high-risk fraud prediction threshold, suspend the network transaction or suspend an account associated with the digital claim.
6 . The non-transitory computer-readable medium as recited in claim 1 , further storing instructions thereon that, when executed by the at least one processor, cause the computing device to:
determine that the fraud prediction does not satisfy a low-risk fraud prediction threshold; and
based on the fraud prediction not satisfying a low-risk fraud prediction threshold:
issue a credit to an account associated with the network transaction; or
provide, for display in the graphical user interface, the visual indicator by providing a low-risk visual indicator for the digital claim.
7 . The non-transitory computer-readable medium as recited in claim 1 , further storing instructions thereon that, when executed by the at least one processor, cause the computing device to:
determine the digital claim disputes an additional network transaction;
identify one or more features associated with the digital claim disputing both the network transaction and the additional network transaction; and
generate the fraud prediction by generating the fraud prediction for the digital claim disputing both the network transaction and the additional network transaction based on the one or more features associated with the digital claim.
8 . The non-transitory computer-readable medium as recited in claim 1 , wherein the fraud detection machine-learning model comprises a gradient boosted decision tree.
9 . The non-transitory computer-readable medium as recited in claim 1 , further storing instructions thereon that, when executed by the at least one processor, cause the computing device to:
determine that the fraud prediction satisfies a high-risk fraud prediction threshold;
based on the fraud prediction satisfying the high-risk fraud prediction threshold, suspend an account associated with the digital claim;
identify an additional digital claim that is currently pending and disputing an additional network transaction associated with the account; and
based on suspending the account and identifying the additional digital claim, provide, for display within the graphical user interface, an additional visual indicator of potential fraud for the additional digital claim.
10 . A system comprising:
at least one processor; and
at least one non-transitory computer-readable medium storing instructions that, when executed by the at least one processor, cause the system to:
receive a digital claim disputing a network transaction;
identify one or more features associated with the digital claim;
generate, utilizing a fraud detection machine-learning model, a fraud prediction for the digital claim based on the one or more features; and
provide, for display in a graphical user interface, a visual indicator of the fraud prediction for the digital claim.
11 . The system as recited in claim 10 , further storing instructions thereon that, when executed by the at least one processor, cause the system to identify the one or more features associated with the digital claim by determining one or more of: an account feature, an automated-clearing-house feature, a computing device feature, a demand draft feature, a transaction feature, a peer-to-peer-payment feature, an identity verification feature, a shared device feature, a shared-internet-protocol-address feature, a customer-service-contact feature, a failed login feature, a password reset feature, a personal identifiable-information-change feature, a linked-claim-dispute feature, a dispute history feature, or a merchant feature.
12 . The system as recited in claim 10 , further storing instructions thereon that, when executed by the at least one processor, cause the system to identify the one or more features associated with the digital claim by determining one or more of: a zip code feature, a merchant category code feature, an account dormancy time feature, a sign in feature, or a transaction-number feature.
13 . The system as recited in claim 10 , further storing instructions thereon that, when executed by the at least one processor, cause the system to:
determine that the fraud prediction satisfies a moderate-risk fraud prediction threshold; and
based on determining that the fraud prediction satisfies the moderate-risk fraud prediction threshold, provide, for display within in the graphical user interface, the visual indicator by providing a moderate-risk visual indicator of fraud for the digital claim.
14 . The system as recited in claim 10 , further storing instructions thereon that, when executed by the at least one processor, cause the system to:
determine that the fraud prediction satisfies a high-risk fraud prediction threshold;
identify a first account associated with the digital claim;
identify one or more features associated with the first account and a second account;
based on the one or more features of the first account and the second account, identify the second account as associated with the first account; and
based on determining that the fraud prediction satisfies a high-risk fraud prediction threshold and identifying the second account is associated with the first account, generate a fraud-propensity label for the second account.
15 . The system as recited in claim 14 , further storing instructions thereon that, when executed by the at least one processor, cause the system to provide, for display within the graphical user interface, a visual fraud propensity indicator for the second account.
16 . The system as recited in claim 14 , further storing instructions thereon that, when executed by the at least one processor, cause the system to identify the one or more features associated with the first account and the second account by determining one or more of: a peer-to-peer-payment feature, a shared device feature, or a shared-internet-protocol-address feature.
17 . A method comprising:
receiving a digital claim disputing a network transaction;
identifying one or more features associated with the digital claim;
generating, utilizing a fraud detection machine-learning model, a fraud prediction for the digital claim based on the one or more features; and
providing, for display in a graphical user interface, a visual indicator of the fraud prediction for the digital claim.
18 . The method as recited in claim 17 , further comprising:
determining that the fraud prediction satisfies a high-risk fraud prediction threshold; and
based on the fraud prediction satisfying the high-risk fraud prediction threshold, suspending the network transaction or suspending an account associated with the digital claim.
19 . The method as recited in claim 17 , further comprising:
determining that the fraud prediction does not satisfy a low-risk fraud prediction threshold; and
based on the fraud prediction not satisfying a low-risk fraud prediction threshold:
issuing a credit to an account associated with the network transaction; or
providing, for display in the graphical user interface, the visual indicator by providing a low-risk visual indicator for the digital claim.
20 . The method as recited in claim 17 , further comprising:
identifying an account associated with the digital claim;
determining the digital claim and an additional digital claim disputing an additional network transaction are associated with the account;
identifying one or more features associated with the additional digital claim; and
generating, utilizing the fraud detection machine-learning model, the fraud prediction by generating the fraud prediction for the digital claim and the additional digital claim based on the one or more features associated with the digital claim and the one or more features associated with the additional digital claim.