Which statement regarding statistical learning methods is true?

The statement regarding statistical learning methods that is true is that when inference is the goal, lasso methods have advantages over bagging methods. Lasso methods, which apply L1 regularization, are particularly effective in producing interpretable models, especially in situations where there are many predictors. They can help identify which predictors are most significant by shrinking the coefficients of less important variables to zero, thereby providing a clearer picture of relationships in the data.

In contrast, bagging methods, such as Random Forests, are primarily aimed at improving prediction accuracy through ensemble learning, which often comes at the cost of interpretability. While bagging can reduce variance and improve predictions, it does not lend itself as easily to inference in terms of understanding the impact of individual predictors.

This distinction highlights how lasso methods are suited for inference-focused contexts, as they maintain a level of interpretability while selecting important features, which is crucial when the goal is to understand relationships in the data rather than just achieve high predictive performance.