Model Training & Adaptation
Decision tree
A prediction method that follows a series of questions, taking a different path depending on each answer.
Example
An AI system asks successive questions about an applicant's supplied features.
Why people use it
It can make predictions through a sequence of relatively easy-to-follow choices.
What you'll hear
“Which branch did this case follow?”
What this means for you
Inspect the actual tree complexity rather than assuming transparency from the AI system name.
Can you control it?
Developer-only
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Can a question tree become difficult to follow?
- Yes. Many branches can make it hard to see how a particular answer was reached.
- Can a small change send a case down a different path?
- Yes. Crossing a cutoff at one question can lead to a different branch and final prediction.
- Can a tree fit past examples too closely?
- Yes. A very detailed tree may copy quirks in old examples that do not help with new cases.