https://communities.sas.com/t5/SAS-Data-Science/SAS-EM-decision-tree-purity/m-p/451344#M6858 <P>Hi,</P><P>&nbsp;</P><P>I have experience in SAS but I'm having trouble using SAS EM. I have a few questions:</P><P>&nbsp;</P><P>1. How is purity measured/ranked?</P><P>2. Is the purity of a node based on the training data or the validation data?&nbsp;</P><P>3. Why is the purity of a node indicated by training/validation (answer to number 2) data?</P><P>4. How do you know (by looking at a tree), which nodes are the purist?</P><P>&nbsp;</P><P>Thanks</P> Wed, 04 Apr 2018 23:05:11 GMT jwoods 2018-04-04T23:05:11Z https://communities.sas.com/t5/SAS-Data-Science/SAS-EM-decision-tree-purity/m-p/451344#M6858 <P>Hi,</P><P>&nbsp;</P><P>I have experience in SAS but I'm having trouble using SAS EM. I have a few questions:</P><P>&nbsp;</P><P>1. How is purity measured/ranked?</P><P>2. Is the purity of a node based on the training data or the validation data?&nbsp;</P><P>3. Why is the purity of a node indicated by training/validation (answer to number 2) data?</P><P>4. How do you know (by looking at a tree), which nodes are the purist?</P><P>&nbsp;</P><P>Thanks</P> Wed, 04 Apr 2018 23:05:11 GMT https://communities.sas.com/t5/SAS-Data-Science/SAS-EM-decision-tree-purity/m-p/451344#M6858 jwoods 2018-04-04T23:05:11Z https://communities.sas.com/t5/SAS-Data-Science/SAS-EM-decision-tree-purity/m-p/452540#M6871 <P>You can use Gini impurity as the splitting criterion when growing the decision tree, which has formula:</P> <P>&nbsp;</P> <P>&nbsp;</P> <P>&nbsp;</P> Mon, 09 Apr 2018 16:01:26 GMT https://communities.sas.com/t5/SAS-Data-Science/SAS-EM-decision-tree-purity/m-p/452540#M6871 WendyCzika 2018-04-09T16:01:26Z