What are the advantages of small decision trees in comparison to larger ones?

What are the advantages of small decision trees in comparison to larger ones?

There are several reasons why small decision trees are preferred. 

• A human expert finds it easy to analyze, explain, and perhaps even correct, a decision tree that consists of no more than a few tests. That is, small trees are easily interpretable. The larger the tree, the more difficult this is.

• Another advantage of small decision trees is their tendency to dispose of irrelevant and redundant information. Whereas the relatively large trees do not support such economy. This advantage of small trees may be helpful in domains where certain attribute values are expensive or time-consuming to obtain.

• Finally, larger trees are prone to overfit the training examples. This is because the divide-and-conquer method keeps splitting the training set into smaller and smaller subsets, the number of these splits being equal to the number of attribute tests in the tree. Ultimately, the resulting training subsets can become so small that the classes may get separated by an attribute that only by chance—or noise—has a different value in the remaining positive and negative examples.

***********************

Related Questions:

• List the advantages and disadvantages of maximum likelihood methods

• What are the properties of Maximum-Likelihood Estimators

Advantages of small decision trees

Compare small and large decision trees

Decision trees and advantages

Why small decision trees are preferable over large decision trees?