SAS Enterprise Miner: Extract predicted values Decision Trees

0
votes

I've been looking for the solution to this for since long. I need to extract the resultant predicted values on running a Decision tree model on Test dataset in SAS Enterprise Miner. However, I see the fit stats, output, treemap etc but couldn't find a way to retrieve the predicted values.

Could somebody please help? Thank You!

1
sasdecision-treepredictionenterprise-miner
'Test' dataset a partition that you created using the partition node, or a different dataset that you've used the Score Node on? Or are you looking for the Score Node? You need to clarify your question. – Reeza
I have used a Data Partition node to create Train (60%), Validation (30%) and Test (10%) partitions. When I run DT model, I see the result only on Train and Validation data set. In the Output window of the DT model, I see Predicted and decision variables. I read somewhere that these can be accessed. Do you know how that can be done? – Ashwini
Somewhere in the node it will say output datasets and let you know what the dataset name is that holds the estimates. If you know the library where the temp files are being saved you can browse there as well. – Reeza
Yeah, it worked the way @Seba explained. But, I don't want to save my predictions and then import the files again. Instead access directly at the output of tree model. Please let me know. – Ashwini
I'm 90% certain the datasets exists already, you just need to know where to find them. If you post on communities.sas.com under Analytics/Data Mining, one of the SAS developers can answer this question. – Reeza

1 Answers

2
votes

There are a number of different ways you can get the values for your tree model out of Enterprise Miner.

The most simple would be to attach a save data node from the utilities menu after the tree node. From there you only specify the output data library where you want your files to be placed.

You should get three output files, one of those will be a file named tree_test that contains a value starting with P_[your target variable], this will be your resultant predictive value.