Whenever you're ready, walk me through how you'd approach this.
What techniques would you use for feature selection?
Explain how you would compare filter, wrapper, and embedded methods, including when each is appropriate. Address correlated features, high-dimensional data, leakage prevention, model interpretability, and how you would validate that selected features improve generalization. Provide Python pseudocode or implementation for a supervised classification dataset.