Conditional Decision Trees

Conditional Decision Trees are a type of Tree-based Model where nodes are split based on specific conditions. Unlike traditional decision trees like C4.5 and C5.0 or Classification and Regression Tree (CART), which split nodes based on the entire dataset, conditional decision trees use conditions that are specific to subsets of the data.

These trees are particularly effective when dealing with high-dimensional or heterogeneous data. Conditional decision trees aim to optimize the splits at each node by using specific conditional criteria, which can lead to better performance in scenarios where standard decision trees might struggle due to the complexity of the data.

Notes:

• Conditional decision trees can deal with datasets that have a mix of continuous, discrete, and categorical variables or when considering different conditions for distinct subsets of the data.
• Local decision trees are implementation of Conditional Decision Trees approach. In this case, decision trees are built specifically for sub-regions of the dataset, allowing for local conditions to be captured accurately.