Hollow-tree super: A directional and scalable approach for feature importance in boosted tree models

• Published in: PloS one Vol. 16; no. 10; p. e0258658
• Main Authors: Doyen, Stephane, Taylor, Hugh, Nicholas, Peter, Crawford, Lewis, Young, Isabella, Sughrue, Michael E.
• Format: Journal Article
• Language: English
• Published: San Francisco Public Library of Science 25.10.2021; Public Library of Science (PLoS)
• Subjects: Algorithms; Analysis; Biology and Life Sciences; Brain; Classification; Computational neuroscience; Computer and Information Sciences; Data analysis; Data visualization; Datasets; Decision making; Decision tree; Decision trees; Engineering and Technology; Learning algorithms; Machine learning; Medical imaging; Medicine and Health Sciences; Mental disorders; Methodology; Methods; Modelling; Neurosciences; Permutations; Physical Sciences; Research and Analysis Methods; Schizophrenia; Social Sciences; Trees; Australia; Sydney New South Wales Australia

Current limitations in methodologies used throughout machine-learning to investigate feature importance in boosted tree modelling prevent the effective scaling to datasets with a large number of features, particularly when one is investigating both the magnitude and directionality of various feature...