1
$\begingroup$

I read this question somewhere(one statement is apparently false) and all four seem to be TRUE to me. What am I missing? What statement is FALSE? (Sharing my thoughts below the statements)

  1. Bagging is used with Decision Trees (This seems obviously True)

  2. Bagging creates random subsamples from the whole dataset with replacement (This is also True)

  3. Each individual predictor is expected to have lower bias than the bagged model. (Individual predictors have low bias or low error and hence high variance, that's why we use bagging so this also seems TRUE)

  4. Aggregation balances bias and variance better than an individual predictor. (This seems TRUE as this is the whole purpose of bagging)

$\endgroup$
1
  • $\begingroup$ Welcome to CV.SE. $\endgroup$
    – usεr11852
    Nov 9, 2022 at 0:53

1 Answer 1

1
$\begingroup$

The third statement is wrong. In a random forest, the bias of a random forest is approximately equal to the bias of any of the individual sampled trees that constitute the forest (but not any large unpruned tree using all the covariates available trained outside the forest, such a tree can indeed have lower bias than the RF). Please see the thread here at: Why does a bagged tree / random forest tree have higher bias than a single decision tree? for more details.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.