1. James
G.
Witten
D.
Hastie
T.
& Tibshirani
R. (2013). An Introduction to Statistical Learning. Springer.
Reference Details: Chapter 8
Section 8.1
"The Basics of Decision Trees
" Page 315. The text states
"Trees are very easy to explain to people. In fact
they are even easier to explain than linear regression! ... This interpretability is a key advantage of tree-based methods." This directly supports the choice of decision trees for their explainability.
2. Amazon Web Services (AWS). (2020). Amazon SageMaker Built-in Algorithms.
Reference Details: In the context of tree-based algorithms like XGBoost
the documentation often highlights their ability to provide feature importance scores
a key aspect of interpretability. While ensemble methods are more complex
the underlying principle of the base learner (the decision tree) being interpretable is a foundational concept in machine learning. The need for services like SageMaker Clarify arises specifically because models like neural networks lack this inherent interpretability.
3. AWS Machine Learning University. (2021). GitHub Repository for Machine Learning University.
Reference Details: In the "Decision Trees and Ensemble Methods" module
the courseware explains that a primary advantage of single decision trees is their high interpretability. The visual representation of the tree provides a clear map of the decision rules used for classification
contrasting with the opaque nature of more complex models.