topic Re: SAS Visual Statistics and Visual Analytics in SAS Visual Analytics

SAS Visual Statistics and Visual Analytics

sachinkalra — Tue, 19 Jun 2018 07:58:15 GMT

I am using SAS VA 9.4.

On a sample dataset I have done some data exploration and want to channel that output to VA dashboard/report.

Also I want to test my model on another test data set. I am unable to find any option relevant to that. Please assist in same

in short:

1.I want to show my predictions/forecast from VS on VA-dashboard and,

2. want to test another sample data on prepared model.

Re: SAS Visual Statistics and Visual Analytics

MikeJenista — Tue, 26 Jun 2018 14:52:20 GMT

(1) This is only partially supported for VA 7.4 (which I think is what you have, the last VA 7.X release but with SAS 9.4). Some objects, like Decision Tree and Forecasting, can be included in reports. But most of the modeling objects (Linear Regression, Clustering, etc) are not supported objects in reports.

(2) You should be able to export the scoring script from your model if it is a VS model (cluster, LinReg, LogReg, GLM, DTree). That script can be run in a SAS session against any data with the expected column names for the model.

Once you have applied the script to a new data set, you can import the results to either VAE or VAD for further analysis or reporting. For VAD, you would only be able to import the scored values as a column in a table, not the details of the model itself.

Re: SAS Visual Statistics and Visual Analytics

sachinkalra — Tue, 17 Jul 2018 07:12:40 GMT

YES.

I was trying to circumvent this whole code interference issue and was actually try to get my model as a node/object as we have in enterprise miner.

The solutions which I have landed upon so far are:

1. Create a data query , call your scoring table and copy paste the score code exported from VS, schedule it and output will keep updating on report on VA.

2. Create a stored process and incorporate that stored process on VA as a object.

Re: SAS Visual Statistics and Visual Analytics

sachinkalra — Tue, 17 Jul 2018 07:14:02 GMT

Thanks Mike.