Decision Trees Overview

These flashcards provide definitions and insights into key concepts related to Decision Trees as discussed in the lecture notes.

Decision Tree

A decision tree is a flowchart-like structure used to classify data into predefined classes based on a set of decision rules. Algorithm can pick the same attribute more than once, don’t use all. stops when only one outcome. Drawbacks: some attributes should be considrered

Root Node

The top attribute in a decision tree from which decisions branch out.

Child Node

Subsequent attributes in a decision tree that follow from the root node.

Leaf Node

The final nodes in a decision tree that contain the values of the class (outcome) variable.

Recursive Partitioning

The process of splitting observations based on attributes to maximize agreement.

Entropy

A measure of the degree of agreement among observations in a given partition; lower entropy indicates higher agreement. lower=more agreement. m=2 for binary, pk is prob. of class

Hyper-parameter

Constraints put on the algorithm that affect the decision tree creation, such as maximum tree depth.

Agreement

The extent to which observations in a given partition have the same value for the outcome variable.

Classification

The process of predicting the target class for a new observation based on decision rules.

Outcome Variable

The variable that represents the result of a classification decision, such as getting a loan (Yes/No).