About WendyCzika

WendyCzika · ‎11-08-2017

Actually I just tried this work-around that worked: put a Model Import node in between HP GLM and Model Comparison, then I think it will include HP GLM for discrete distributions in your Model Comparison.

WendyCzika · ‎11-08-2017

Poisson, Negative binomial, and the ZI versions of those are the discrete ones.

WendyCzika · ‎11-08-2017

When you specify a discrete distribution for the Interval Target Probability Distribution property, model assessment is not performed and model comparison with the Model Comparison node is unavailable.The Fit Statistics table from the HPGENSELECT procedure shown in the Output window can help you compare between 2 HP GLM runs.

WendyCzika · ‎11-07-2017

In either the Input Data node or a Metadata node inserted in your flow before Decision Tree, you can change the level of that variable to Interval to treat as continuous or to set the role to Rejected to exclude it from the model.

WendyCzika · ‎11-06-2017

In HPNEURAL, you use INPUT statements to define your inputs, with the appropriate LEVEL= option, e.g.: input classvar1 classvar2 .... / level=nom; input continuousvar1 continuousvar2 ... / level=int;

WendyCzika · ‎11-03-2017

The bins (segments) are based on quantile binning of your predicted target (P_target).

WendyCzika · ‎11-02-2017

It can represent either clusters from the Clustering node or the leaf from a Decision Tree. There might be other nodes that produce it, but those are probably the two most common.

WendyCzika · ‎11-02-2017

There are instructions and a sample SAS Enterprise Miner diagram (the XML file) at this GitHub repo: https://github.com/sassoftware/enlighten-integration/tree/master/SAS_EM_PythonIntegration This shows how you can run your Python code from a SAS Code node, then use the Model Import node to be able to assess your model as you can with SAS Enterprise Miner models.

WendyCzika · ‎11-01-2017

If you want to compare all of the individual models to the ensemble model, then you would connect all 4 (NN, Log Reg, DTree, Ensemble) to Model Comparison.

WendyCzika · ‎11-01-2017

Typically holdout and test mean the same thing - since the validation partition can be used in training your model (for pruning, early stopping, etc.), you keep the test or holdout for final assessment, based on data not used in any part of training your model.

WendyCzika · ‎10-27-2017

Does the Input Data node for the data you are trying to score have Role set to Score?

WendyCzika · ‎10-25-2017

You can use as many Decision Tree nodes as you'd like with different properties, then connect them to a Model Comparison node to see which one performs the best, see attached image.

WendyCzika · ‎10-25-2017

If your other data set has the target in it, and you are using it as a hold-out data set for assessing your model, then the role should be Test. If your data does not contain the target and you want to score it to get predictions, then the role should be Score.

WendyCzika · ‎10-19-2017

The EM Reference Help has this information, examples, etc.

WendyCzika · ‎10-16-2017

When you say you can't access it, do you mean the property is disabled or you get an error when you try to open the Interactive window? The first would be caused by having either User Frozen Tree or Import Tree Model set to Yes, the second could be caused by not having run the predecessor nodes.

Online Status	Offline
Date Last Visited	3 weeks ago