topic Re: SAS Enterprise Miner - Decision Tree in SAS Data Science

SAS Enterprise Miner - Decision Tree

okbbachabi0 — Sun, 05 Nov 2017 16:31:02 GMT

Hi Guys

I'm too new to SAS.

I'm working on a project predicting a model.

I have designed my diagram as shown on the attachment.

When i run the program at the decision tree node I'm getting a run time error.

So i checked on the result logs (see attachement) and found that the variable has more than 32766 categorical values.

I dont know how to go about this. Any help will be appreciated.

Thanks guys.

Re: SAS Enterprise Miner - Decision Tree

HEOR-Ryan — Mon, 06 Nov 2017 14:54:50 GMT

I would suggest changing the variable to a numeric or continuous measure. It is unlikely to be categorical with that many levels.

Right right-click the dataset and select edit variables to change the type

Re: SAS Enterprise Miner - Decision Tree

okbbachabi0 — Tue, 07 Nov 2017 19:46:54 GMT

Hi HEOR-Ryan

Thanks for the reply.

Do you mean to right click on the actual data set as that doesn't give me the option to edit the variables even?

How can I go about this conversion?

thanks

Re: SAS Enterprise Miner - Decision Tree

WendyCzika — Tue, 07 Nov 2017 20:14:59 GMT

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.

Re: SAS Enterprise Miner - Decision Tree

okbbachabi0 — Thu, 09 Nov 2017 13:08:52 GMT

Hi WendyCzika

Thanks. I have tried what you said but when i clicked on the file import node to change the level. The interval is greyed out. From the metadata when I added to the flow doesn't allow me to change the level. I don't want to reject those variables as i need them for my prediction.

Any other advices?

Thanks

Re: SAS Enterprise Miner - Decision Tree

MikeStockstill — Thu, 09 Nov 2017 15:27:38 GMT

Hello okbbachabi0 -

The Interval selection is grayed out if the variable is defined as character in the underlying SAS data set. The Interval selection is available only for numeric variables.

Solutions include:

- convert the character variable to a numeric variable

- return to the method that you used to create the SAS data set, and define the variable as numeric instead of character.

Have a great day.

Re: SAS Enterprise Miner - Decision Tree

okbbachabi0 — Sat, 11 Nov 2017 20:24:45 GMT

MikeStockstill

Thanks buddy.

Working perfectly.

Great job.