I have trained a “gradient boosted tree model” to my dataset, which works surprisingly really good. Now I would like to analyse the model to find out, which data input (my model uses 24 input columns, one target column) has the biggest influence on the result.
Is there a node to do such a sensitivity analysis available? How do you check your models?
Hello @ActionAndi -
The standard Gradient Boosted Trees Learner node currently doesn’t provide a measure of variable importance. The H2O Gradient Boosting Machine Learner, however, does!
Check out this workflow:
Wow, that is exactly the node I was looking for!
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.