Explanation

Statement I is correct: CART does not require specifying an initial hyperparameter like K in KNN. While CART has parameters like tree depth, it doesn't have a mandatory hyperparameter equivalent to K.

Statement II is correct: CART does not require specifying a similarity or distance measure, unlike KNN which relies on distance metrics (Euclidean, Manhattan, etc.) to find nearest neighbors.

Statement III is correct: CART provides a visual explanation through its tree structure, making it interpretable and allowing users to understand the decision path, while KNN is more of a "black box" approach.

All three statements accurately describe advantages of CART over KNN.