Breakdown of algorithm

Breakdown of algorithm

• Divide and Conquer

• Given a training set $D_T$ with $M$ classes and a homogeneity measure.

• Recursively split the set into subsets with less homogeneity until all subsets are homogeneous.

• recursively dividing the dataset based on feature tests until it creates a tree structure that represents the underlying patterns in the data.

• |D|: Represents the number of instances in the dataset $D$

• $\forall \mathbf{x} \in D$ : For all instances $x$ in the dataset $D$.

• $y_m$ : Represents a class label.

• Return leaf with default class: If the dataset is empty, it returns a leaf node with a default class.

• Return leaf with class label $y_m$, containing D: If the dataset is homogeneous (all instances have the same class), it returns a leaf node with the class label and includes the dataset $D$ in that leaf.

• Select a test based on a single input variable: This step involves selecting a feature and a corresponding condition to split the dataset into subsets.

• Split D into $D_1, D_2, \ldots, D_O$: The dataset $D$ is split into subsets  $D_1, D_2, ..., D_O$  based on the selected test, where $O$ is the number of outcomes.

• for $O=1$ to 0 do INDUCETREE($D_o$): Recursively apply the  INDUCETREE  function to each subset $D_o$ .