Hi Funda, many thanks for providing this detailed flow! I replicated it for the simple decision tree because the property "Cross Validation" of the Decision Tee node is a bit of a black box for me. Below you find my results: The _ASE_ of the Model Import node is equal to the _ASE_ of the End Group node where Group = _OVERALL_. The _ASE_ of the _OVERALL_ Group however is calculated in accordance with the Flow Code which is for a 3-fold cross validation: if ^(_fold_ =1) then do; ...*prediction based on the tree trained on fold 2&3* end; if ^(_fold_ =2) then do; ...*prediction based on the tree trained on fold 1&3* end; if ^(_fold_ =3) then do; ...*prediction based on the tree trained on fold 1&2* end; But shouldn't be the code in the following way to obtain the precitions on the holdout samples?: if (_fold_ =1) then do; ...*prediction based on the tree trained on fold 2&3* end; if (_fold_ =2) then do; ...*prediction based on the tree trained on fold 1&3* end; if (_fold_ =3) then do; ...*prediction based on the tree trained on fold 1&2* end; I am probably missing something here...Please let me know in case I can provide more information. Many thanks in advance!
... View more
