Quick Answer

The answer is the chi-squared test and mutual information. These two methods are appropriate for feature selection during exploratory data analysis because they both measure the dependency between each feature and the target variable, allowing you to rank features by predictive power. The chi-squared test is specifically designed for categorical features and a categorical target, making it a natural fit when the target is binary, while mutual information is more flexible as it can handle both categorical and continuous features by quantifying the amount of information shared between variables. On the AWS Certified Machine Learning Specialty MLS-C01 exam, this question tests your ability to distinguish between supervised and unsupervised techniques during EDA—a common trap is confusing dimensionality reduction methods like PCA, which are unsupervised and ignore the target, with supervised filters like chi-squared and mutual information. Remember that PCA and k-means are unsupervised, while chi-squared and mutual information directly use the target to select features. A helpful memory tip: if you need to "inform" your model about the target, use mutual information; if you need to "test" categorical associations, use chi-squared.