Question on GDBT : perfect prediction?

Module 6. Ensemble of models Ensemble based on boosting
#1

Hi,

In the GDBT section, it is said that the prediction for the residual is perfect for the orange data point, but I see there is a gap.

temp-scikit

Could you explain how to analyse, observe this perfect fit on the curve ?

Also, the residual error prediction is :
Prediction of the residual for x=-0.382: 0.264

But it is different from the tree error above :
Error of the tree: 0.383

But it is said they are the same

I think I didn’t understand very well

Thanks in advance for the answer :slight_smile:

1 Like
#2

Could you synchronize your notebook (File → Revert to original). We did correct this mistake last week because the sample chosen was indeed not fitting with the discussion.

In case, that you want to compare your version of the notebook, you should get the same as: Gradient-boosting decision tree (GBDT) — Scikit-learn course

Basically you should get the following image:

image

In this case, you can see that there is not residuals on the orange dot sample.