Cross validation with decision trees
Hello everyone, when using for example a k fold cross validation strategy with 10 folds with decision trees, 10 trees are built.
When outputting the results for cross validation, only one tree/model is shown. How is that tree calculated exactly?
When outputting the results for cross validation, only one tree/model is shown. How is that tree calculated exactly?
Find more posts tagged with
Sort by:
1 - 2 of
21
Thanks, it's much clearer now!
Sort by:
1 - 1 of
11
Hi!
When the model output of the Cross Validation operator is connected, it calculates a last (11th) model from all the data. So your tree was built from all data and applying the training phase on them.
If you're interested in the ten other trees, you can set a breakpoint inside the cross validation and take a look at them. With some automation (e. g. Generate Macro incrementing the loop number) you can also store the validation models in the repository.
Regards,
Balázs
When the model output of the Cross Validation operator is connected, it calculates a last (11th) model from all the data. So your tree was built from all data and applying the training phase on them.
If you're interested in the ten other trees, you can set a breakpoint inside the cross validation and take a look at them. With some automation (e. g. Generate Macro incrementing the loop number) you can also store the validation models in the repository.
Regards,
Balázs
When the model output of the Cross Validation operator is connected, it calculates a last (11th) model from all the data. So your tree was built from all data and applying the training phase on them.
If you're interested in the ten other trees, you can set a breakpoint inside the cross validation and take a look at them. With some automation (e. g. Generate Macro incrementing the loop number) you can also store the validation models in the repository.
Regards,
Balázs