A good estimator is one for which the test error is smallest.
High bias: high train error ~= high test error
High variance: low train error << high test error
High bias: high train error ~= high test error
High variance: low train error << high test error
No comments:
Post a Comment