Hi!
How does predictor importance calculate in SPSS Modeler? ( It is mentioned with the help of SW that the predictor importance is relative, So, it is not based on OR?)
In different models of DT such as C5.0 the branches of a tree are not according to the results of predictor importance. I mean the root is not the most important predictor followed by the next tree layers! why?
Also, how does predictor importance work in boosting?
The pic of the tree branches is attached.
Your question is not well-clear but you can find more here www.Stats4Edu.com/services
Relative means the sum of importance values for all predictors is 1. Hence an input A with 0.4 importance value is twice as important as Input B with an important value of 0.2. There is little docu but this is how I read it...
- Badges
- Science topic
- Similar topics
- Filing