Abstract
This research presents Gradient Boosted Tree High Importance Path Snippets (gbt-HIPS), a novel, heuristic method for explaining gradient boosted tree (GBT) classification models by extracting a single classification rule (CR) from the ensemble of decision trees that make up the GBT model. This CR contains the most statistically important boundary values of the input space as antecedent terms. The CR represents a hyper-rectangle of the input space inside which the GBT model is, very reliably, classifying all instances with the same class label as the explanandum instance. In a benchmark test using nine data sets and five competing state-of-the-art methods, gbt-HIPS offered the best trade-off between coverage (0.16–0.75) and precision (0.85–0.98). Unlike competing methods, gbt-HIPS is also demonstrably guarded against under- and over-fitting. A further distinguishing feature of our method is that, unlike much prior work, our explanations also provide counterfactual detail in accordance with widely accepted recommendations for what makes a good explanation.
Highlights
Data Analytics and Artificial Intelligence Research Group, Faculty of Computing, Engineering and the Built Environment, Birmingham City University, Curzon Street, Birmingham B5 5JU, UK; Abstract: This research presents Gradient Boosted Tree High Importance Path Snippets, a novel, heuristic method for explaining gradient boosted tree (GBT) classification models by extracting a single classification rule (CR) from the ensemble of decision trees that make up the GBT model
Such tasks are often still found in high-stakes decision making domains, such as medical decision making [5,6,7,8]; justice and law [9,10]; financial services [11,12,13]; and defence and military intelligence [14]. In these and similar domains, there is a high burden of accountability for decision makers to explain the reasoning behind their decisions. This burden only increases with the introduction of machine learning (ML) into decision making processes [15]
interpretable machine learning (IML) methods can be used to facilitate the interpretation of a GBT model, as well as other types of decision tree ensemble, known as decision forests (DFs)
Summary
Data Analytics and Artificial Intelligence Research Group, Faculty of Computing, Engineering and the Built Environment, Birmingham City University, Curzon Street, Birmingham B5 5JU, UK; Abstract: This research presents Gradient Boosted Tree High Importance Path Snippets (gbt-HIPS), a novel, heuristic method for explaining gradient boosted tree (GBT) classification models by extracting a single classification rule (CR) from the ensemble of decision trees that make up the GBT model. A further distinguishing feature of our method is that, unlike much prior work, our explanations provide counterfactual detail in accordance with widely accepted recommendations for what makes a good explanation In these and similar domains, there is a high burden of accountability for decision makers to explain the reasoning behind their decisions. IML methods can be used to facilitate the interpretation of a GBT model, as well as other types of decision tree ensemble, known as decision forests (DFs). These methods generate a cascading rule list (CRL) as an inherently interpretable proxy model.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
