US 12,423,619 B2
Class-specific predictor importance
Lei Tian, Xi'an (CN); Han Zhang, Xi'an (CN); Ning Zhang, Xi'an (CN); Na Zhao, Xi'an (CN); Li Na Wang, Xi'an (CN); and Rong Hong Wan, Xi'an (CN)
Assigned to International Business Machines Corporation, Armonk, NY (US)
Filed by INTERNATIONAL BUSINESS MACHINES CORPORATION, Armonk, NY (US)
Filed on Mar. 31, 2022, as Appl. No. 17/709,646.
Prior Publication US 2023/0316125 A1, Oct. 5, 2023
Int. Cl. G06N 20/00 (2019.01); G06F 16/22 (2019.01); G06N 5/02 (2023.01)
CPC G06N 20/00 (2019.01) [G06F 16/2246 (2019.01); G06N 5/02 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method within a computer hardware system including a machine learning engine and a predictor indicator importance generator, comprising:
generating, using the machine learning engine, a prediction model having
a tree structure including a plurality of nodes,
a plurality of class label subgroups, and
a plurality of class-specific predictor indicators;
causing a first display, in a graphical user interface interacting with the computer hardware system, of a representation of the tree structure including the plurality of class label subgroups and the plurality of class-specific predictor indicators;
receiving, based upon a user input, a selected class label subgroup from the displayed plurality of class label subgroups and a selected class-specific predictor indicator from the displayed plurality of class-specific predictor indicators;
generating, using the predictor indicator importance generator, a class-specific predictor indicator importance for the selected class-specific predictor indicator by merging a term predictor indicator frequency and a purity predictor indicator frequency at each instance of the selected class-specific predictor indicator within one or more nodes of the tree structure; and
causing a second display, in the graphical user interface, of the representation of the tree structure, wherein each path between a root node of the tree structure and a leaf node mapped to the selected class label subgroup having a node containing an instance of the selected class-specific predictor indicator is depicted in a manner that contrasts with other paths between the root node and other leaf nodes of the tree structure.