Machine learning models for evaluating entities in a high-volume computer network
In an example, a machine learning algorithm is used to train an entity risk evaluation model to output an entity risk score based on transaction data in a computer network. Entity risk scores for various entities may be stored in a database, and retrieved and displayed upon user interaction with one or more reports involving corresponding entities.
1. A system comprising:
at least one hardware processor;
a computer-readable medium storing instructions, which when executed by the at least one hardware processor, performs operations comprising:
obtaining a non-training set of transaction data from a database, each piece of the non-training set of transaction data including an identification of a supplier entity and a receiver entity;
extracting one or more features from the non-training set of transaction data;
generating one or more key performance indices from the one or more features, wherein the key performance indices are measurements calculated from the one or more features;
feeding the one or more key performance indices into an entity risk evaluation model to generate an entity risk score for each of a plurality of supplier entities and receiver entities identified in the non-training set of transaction data, the entity risk evaluation model having been trained by obtaining a training set of transaction data from the database, the training set of transaction data including an identification of a supplier entity and a receiver entity, extracting one or more features from the training set of transaction data, generating one or more key performance indices from the one or more features from the training set, and feeding the one or more key performance indices from the training set into a machine learning algorithm to train the entity risk evaluation model to generate an entity risk score for each entity identified in the transaction data;
storing the entity risk scores for each of the plurality of supplier entities and receiver entities in the database;
retrieving the entity risk score for a first entity from the database; and
causing display of the entity risk score for the first entity in a graphical user interface.
2. The system of claim 1 , wherein the feeding the one or more key performance indices into the machine learning algorithm to train the entity risk evaluation model to generate an entity risk score for each entity identified in the transaction data includes learning a different weight to be applied to each category of key performance index.
3. The system of claim 2 , wherein the entity risk evaluation model includes a model trained via a scorecard-based approach.
4. The system of claim 2 , wherein the entity risk evaluation model includes a model trained via an unsupervised approach.
5. The system of claim 2 , wherein the entity risk evaluation model includes a model trained via a semi-supervised approach.
6. The system of claim 2 , wherein the entity risk evaluation model includes a model trained via a supervised approach.
7. The system of claim 2 , wherein the entity risk evaluation model is based on a weighted average of outputs of a model trained via a scorecard-based approach, a model trained via an unsupervised approach, a model trained via a semi-supervised approach, and a model trained via a supervised approach.
8. A method comprising:
obtaining a non-training set of transaction data from a database, each piece of the non-training set of transaction data including an identification of a supplier entity and a receiver entity;
extracting one or more features from the non-training set of transaction data;
generating one or more key performance indices from the one or more features, wherein the key performance indices are measurements calculated from the one or more features;
feeding the one or more key performance indices into an entity risk evaluation model to generate an entity risk score for each of a plurality of supplier entities and receiver entities identified in the non-training set of transaction data, the entity risk evaluation model having been trained by obtaining a training set of transaction data from the database, the training set of transaction data including an identification of a supplier entity and a receiver entity, extracting one or more features from the training set of transaction data, generating one or more key performance indices from the one or more features from the training set, and feeding the one or more key performance indices from the training set into a machine learning algorithm to train the entity risk evaluation model to generate an entity risk score for each entity identified in the transaction data;
storing the entity risk scores for each of the plurality of supplier entities and receiver entities in the database;
retrieving the entity risk score for a first entity from the database; and
causing display of the entity risk score for the first entity in a graphical user interface.
9. The method of claim 8 , wherein the feeding the one or more key performance indices into the machine learning algorithm to train the entity risk evaluation model to generate an entity risk score for each entity identified in the transaction data includes learning a different weight to be applied to each category of key performance index.
10. The method of claim 9 , wherein the entity risk evaluation model includes a model trained via a scorecard-based approach.
11. The method of claim 9 , wherein the entity risk evaluation model includes a model trained via an unsupervised approach.
12. The method of claim 9 , wherein the entity risk evaluation model includes a model trained via a semi-supervised approach.
13. The method of claim 9 , wherein the entity risk evaluation model includes a model trained via a supervised approach.
14. The method of claim 9 , wherein the entity risk evaluation model is based on a weighted average of outputs of a model trained via a scorecard-based approach, a model trained via an unsupervised approach, a model trained via a semi-supervised approach, and a model trained via a supervised approach.
15. A non-transitory machine-readable storage medium comprising instructions, which when implemented by one or more machines, cause the one or more machines to perform operations comprising:
obtaining a non-training set of transaction data from a database, each piece of the non-training set of transaction data including an identification of a supplier entity and a receiver entity;
extracting one or more features from the non-training set of transaction data;
generating one or more key performance indices from the one or more features, wherein the key performance indices are measurements calculated from the one or more features;
feeding the one or more key performance indices into an entity risk evaluation model to generate an entity risk score for each of a plurality of supplier entities and receiver entities identified in the non-training set of transaction data, the entity risk evaluation model having been trained by obtaining a training set of transaction data from the database, the training set of transaction data including an identification of a supplier entity and a receiver entity, extracting one or more features from the training set of transaction data, generating one or more key performance indices from the one or more features from the training set, and feeding the one or more key performance indices from the training set into a machine learning algorithm to train the entity risk evaluation model to generate an entity risk score for each entity identified in the transaction data;
storing the entity risk scores for each of the plurality of supplier entities and receiver entities in the database;
retrieving the entity risk score for a first entity from the database; and
causing display of the entity risk score for the first entity in a graphical user interface.
16. The non-transitory machine-readable storage medium of claim 15 , wherein the feeding the one or more key performance indices into the machine learning algorithm to train the entity risk evaluation model to generate an entity risk score for each entity identified in the transaction data includes learning a different weight to be applied to each category of key performance index.
17. The non-transitory machine-readable storage medium of claim 16 , wherein the entity risk evaluation model includes a model trained via a scorecard-based approach.
18. The non-transitory machine-readable storage medium of claim 16 , wherein the entity risk evaluation model includes a model trained via an unsupervised approach.
19. The non-transitory machine-readable storage medium of claim 16 , wherein the entity risk evaluation model includes a model trained via a semi-supervised approach.
20. The non-transitory machine-readable storage medium of claim 16 , wherein the entity risk evaluation model includes a model trained via a supervised approach.