Chapter 10. Decision Tree Models
The previous chapter provided an overview of the various types of data mining models (predictive, clustering, and association) that can be developed in Modeler. Predictive modeling is the most common form of data mining, and as was mentioned in the previous chapter, three very different strategies can be employed: statistical, decision tree, or machine learning. In this chapter, we will discuss:
- Decision tree theory
- CHAID theory
- Partition node
- CHAID dialog options
- CHAID results
Before we begin discussing decision tree theory, we'll look at a brief overview of where the next two chapters are going. In this chapter and the next, we are going to build a classification model using Chi-square Automatic Interaction Detection (CHAID). We will then assess its ability to make effective predictions, and finally use the model to score new cases. Our teaching experience has shown us that in order to limit confusion, it is a good idea to briefly revisit the notion of an algorithm...
