US 12,393,877 B2
Attributing reasons to predictive model scores with local mutual information
Matthew Bochner Kennel, San Diego, CA (US); and Scott Michael Zoldi, San Diego, CA (US)
Assigned to Fair Isaac Corporation, Minneapolis, MN (US)
Filed by FAIR ISAAC CORPORATION, Minneapolis, MN (US)
Filed on Nov. 9, 2023, as Appl. No. 18/506,031.
Application 18/506,031 is a continuation of application No. 16/700,982, filed on Dec. 2, 2019, granted, now 11,875,232.
Prior Publication US 2024/0078475 A1, Mar. 7, 2024
Int. Cl. G06N 20/00 (2019.01); G06F 17/18 (2006.01)
CPC G06N 20/00 (2019.01) [G06F 17/18 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A system for providing insights about a machine learning model, the system comprising one or more processors configured for:
using training data to train the machine learning model to learn patterns to determine whether data associated with an event provides an indication that the event belongs to a certain class from among a plurality of classes;
evaluating one or more outputs of the machine learning model to produce a data set pairing observed scores S and computing a set of predictive input variables Vi related to the input features of the machine learning model, the data set not identical to the input features of the machine learning model; and
constructing at least one data-driven estimator based on an explanatory statistic associated with the predictive input variables Vi, packaged with the machine learning model, and utilized to provide a definition of explainability for a score generated by the machine learning model,
the explanatory statistic being a variable relevance statistic (VRS) between the score S and the input variables Vi evaluated as a means of indicating meaningful explanatory variable relationships used by the machine learning model for the generated score and the VRS quantifying the degree of co-occurrence between observed values of the score S and one or more variables Vi; and
performing one or more of a deduplication procedure or an explanatory elucidation procedure to enhance palatability and relevance of the definition of explainability for the score generated by the machine learning model.