| CPC G06N 5/045 (2013.01) [G06N 5/01 (2023.01); G06N 20/00 (2019.01)] | 20 Claims |
|
13. A system comprising:
a data processor; and
memory storing instructions stored on the data processor, which when executed result in operations comprising:
transmitting, from an application to a scoring equation service, a request comprising training data for machine learning (ML) components, the training data comprising one or more scoring equations;
receiving, by the application, a received trained ML model from the scoring equation service based on the training data for the ML components responsive to the request;
generating a user interface (UI);
displaying, in the UI, a list of a plurality of trained machine learning (ML) models comprising at least the received trained ML model;
receiving and processing a user selection, via the UI, of a particular trained ML model from the list of the plurality of trained ML models displayed in the UI;
receiving, by the application, data comprising a specification which defines the particular trained ML model selected by the user;
parsing, by the application, a model description of the particular trained ML model;
creating, by an engine factory, an instance of an engine based on the model description after processing the received user selection of the particular trained ML model;
displaying, in the UI, a request for a prediction and an associated explanation using the instance of engine;
receiving, by the UI, user input data comprising a requested ML prediction and influencer values of two or more influencers;
requesting, by the application, model information associated with the specification, the model information comprising at least one of a model type, a target name, or a target type;
requesting, by the application, model influencers associated with the specification, the model influencers comprising at least one of a name, a value type, a storage type, or a listing of values;
determining and displaying, by the instance of the engine, the requested prediction and the associated explanation based on the model information and the model influencers, wherein the displaying of the requested prediction and the model influencers comprises displaying importance values for each of the two or more influencers,
wherein the prediction and the associated explanation comprises an array of individual contributions associated with each of the two or more influencers, the array comprising an influencer name and importance value computed for the influencer; and
displaying, in the UI, shorter feature names representing influencer names that are mapped to the two or more influencers.
|