topic Re: can I dump enterprise miner graph to sas base code? in SAS Data Science

can I dump enterprise miner graph to sas base code?

cfjdoedens — Tue, 12 Jun 2018 12:44:13 GMT

Hi!

This is about how to get logic from Enterprise Miner into SAS base code.

I know it is possible to train and test in Enterprise Miner for example a support vector machine on some data set, and as a result have a trained recogniser. This recogniser can then be dumped as SAS base code, and used in SAS base (if one has the licenses for the used SAS procs).

Now is it possible, to build a complete graph in Enterprise Miner, that involves all the logic for the training and testing, and dump that into SAS base?

So the result would be SAS base code that can be fed a training data set, which then produces a trained recogniser.

Thanks for you answers!

Background:

My goal is to automatically build hundreds of different recognisers, in a standard way, from hundreds of different training sets, so I want to automate the process of producing recognisers, by coding it in SAS base. If I can not dump the graph from Enterprise Miner I will have to build it by hand in SAS base, using the right procs, and appropriate SAS base code. Would give me better control and cleaner code management anyway.

Re: can I dump enterprise miner graph to sas base code?

WendyCzika — Thu, 14 Jun 2018 23:06:09 GMT

If you are asking about getting the accumulated score code from your Enterprise Miner flow, you can do that using the Score node. See http://go.documentation.sas.com/?docsetId=emref&docsetTarget=n18m9qyz7h180kn1qnkhkjgwzu2e.htm&docsetVersion=14.3&locale=en for more information.

Re: can I dump enterprise miner graph to sas base code?

cfjdoedens — Fri, 15 Jun 2018 08:20:06 GMT

Hi Wendy.

Thanks for your answer!

However, no that is not what I want to accomplish.

I want to build a miner project, a graph so to speak, and dump it into SAS base code.

Then feed the dumped SAS base code my training data set, and then I expect as a result of that a trained recogniser, which I can feed samples to score.

So when I feed the dumped miner project another traning data set, a different recogniser with different parameters might result.

Greetings, Crist-Jan