Answer: Permutation Feature Importance

The question asks for a feature extraction method. Feature extraction involves transforming or combining existing features to create new, more informative ones, or selecting a subset of features. Among the options: A) Mutual information is a feature selection method that measures dependency between variables and can be used to select features but is not typically considered a feature extraction method that creates new features. B) Mood's median test is a non-parametric test for comparing medians of two or more groups, not a feature extraction method. C) Kendall correlation is a rank-based correlation measure used for feature selection by assessing relationships between features and the target variable, but it does not create new features. D) Permutation Feature Importance is a model-agnostic method for evaluating feature importance by shuffling feature values and measuring performance impact; it is used for feature selection, not extraction. The community discussion shows conflicting opinions: one comment with 8 upvotes supports D (Permutation Feature Importance), another with 4 upvotes supports C (Kendall correlation), and a third with 1 upvote supports A (Mutual information). However, based on Azure Machine Learning documentation and best practices, Permutation Feature Importance is explicitly mentioned in Azure ML modules for feature importance assessment, making it the most relevant for an Azure ML task. It is commonly used in feature engineering workflows to identify significant features, though strictly it is a selection method. Given the context, D is the optimal choice as it aligns with Azure ML's provided metrics for performance measurement in feature importance processes.

Author: LeetQuiz Editorial Team