Consider decision tree A learned with min_samples_leaf = 500. Now consider decision tree B trained on the same dataset and parameters, except that the min_samples_leaf=50. Which of the following is/are always true?
The depth of B >= the depth of A
The number of nodes in B >= the number of nodes in A
The test error of B <= the test error of A
The training error of B <= the training error of A
Answers
Answer:
The depth of B >= the depth of A
The number of nodes in B >= the number of nodes in A
The training error of B <= the training error of A
Explanation:
min_samples_leaf guarantees a minimum number of samples in a leaf. Higher no of this parameter means you are stopping early. A lower value allows you to grow further
min_samples_leaf guarantees a minimum number of samples in a leaf. Higher no of this parameter means you are stopping early. A lower value allows you to grow further. As the tree grows no of nodes increases.
min_samples_leaf guarantees a minimum number of samples in a leaf. Higher no of this parameter means you are stopping early. A lower value allows you to grow further. As the tree grows no of nodes increases. With more nodes and deeper tree , it tends to memorize training data and variance of the model increases.
Answer:
A and
Explanation: