From the course: Machine Learning and AI Foundations: Decision Trees with SPSS

Unlock the full course today

Join today to access over 24,700 courses taught by industry experts.

Stopping rules in CHAID and C&RT

Stopping rules in CHAID and C&RT

From the course: Machine Learning and AI Foundations: Decision Trees with SPSS

Start my 1-month free trial Buy for my team

Stopping rules in CHAID and C&RT

“

- [Narrator] CHAID and C&RT both have three important stopping rules. What do we mean by stopping rules? Just quite simply, what makes the tree come to a halt and complete. So we've built the trees, how might we change some settings to alter the trees, and improve them? The first stopping rule is essentially identical, in CHAID and C&RT. It's the parent, child, sample size limits. Let me show you in the modeler software. I'm in the CHAID and C&RT string and we'll start looking at the CHAID. In the stopping rules tab, we can see that we can actually set these settings by percentage or absolute value. Let's start by talking about absolute value. And of course software will differ a little bit in how you control this. In my experience, setting by absolute value will be somewhat more common than setting by percentage. But it's referring to the same thing. Minimum records in a parent branch of 100 means, that unless a node has at least 100 cases, in the case of Titanic that would be 100…

Contents