kaz-Anova / StackNet

StackNet is a computational, scalable and analytical Meta modelling framework
MIT License
1.32k stars 344 forks source link

How to interpret the StackNet program output? #17

Closed ajing closed 7 years ago

ajing commented 7 years ago

I wonder whether the output can give some hint about whether I select the right base models.

I saw something like:

 Total ranks after target inclusion: 212409
 Gain from before : 152317
 percentage of unique ranks versus elements size: 13.892484314388511%
 Gain of percentage of unique ranks versus elements size: 9.962202794207%

What does it mean? Are they for monitoring?

Thanks

kaz-Anova commented 7 years ago

This is is presented if you leave verbose=true and just shows you how good the compression ratio is when building the histogram in tree-based methods. If any it just shows how fast it will be (relative to other parameters) and it is not very helpful to gauge performance.

Your best bet is to tune one parameter at a time using cross validation .