How Do I Move From SAS® Enterprise Miner™ to SAS® Viya®? Q&A, Slides, and On-Demand Recording

[#1]

Watch this Ask the Expert session to learn how to move from SAS Enterprise Miner to SAS Viya. 

Watch the webinar

Join expert Melodie Rush as she shows you how your SAS Enterprise Miner models can easily be brought into Viya pipelines both to compare with new models and to score new data. During this webinar you will learn:

• Differences and similarities between SAS Enterprise Miner and SAS Visual Data Mining and Machine Learning.
• How to use SAS Enterprise Miner models in Viya and Viya models in SAS Enterprise Miner.
• Scoring in Viya with SAS Enterprise Miner models and scoring in SAS 9.4 with Viya models.
• Other functionality.

The questions from the Q&A segment held at the end of the webinar are listed below and the slides from the webinar are attached.

Q&A

Is the gradient boosting algorithm in EM the same as the one in VDMML?

It’s a different implementation of gradient boosting. Because of the architecture in VDMML, we’ve been able to implement a richer set of options with gradient boosting. If you run one in EM, you would get similar, but not the exact same results in VDMML. The settings are different, and you have different options for optimization.

Do I need the same dataset that I used to create the model in SAS Enterprise Miner to use the model in Viya Pipelines?

You don’t have to have the same data set, but you do have to have the same data structure. If your original has 5 columns or variables, you need those same 5 variables, and they need to be in the same structure (same column names for the data).

How do you do a data partition in SAS Viya VDMML?

Data partitioning is available in the visual interfaces Model Studio and Explore and Visualize. It is also available in SAS Studio as a task. For Explore and Visualize, you can add a partition variable using the New Data Item in the data pane. For Model Studio the data is automatically partitioned when you create the project, you can indicate a variable to have the role of partition.

SAS Viya has a Data Preparation component available to all products on Viya. That component has tasks to allow you to partition your data before bringing that data set into VDMML. This documentation has information about partitioning using a task in SAS Studio (Develop SAS Code).

There was a programming guide for Enterprise Miner. Is there an equivalent guide for VDMML?

Yes, we have the equivalent for VDMML.

You may also find these helpful:

Procedures

Action Sets

Open Source (Python)
Open Source (R)

GitHub

Can I export the models I develop and the supporting SAS runtime from Viya/VDDML so that I can run them locally in my own cloud-based solution, or must I call my models remotely and get the results back using a VDMML URL?

It depends on how you export your models and what you have in your cloud-based solution. If you export a data step, then that’s SAS and you would have to have SAS in your cloud-based solution so you can run it. If you export it as a REST API and you have some type of programming language that can run that REST API, then you can run it in your cloud-based solution. It depends on your architecture, structure and what you have available.

Previously in EM you could have multiple model nodes and manually select your best model based on business criteria rather than only statistical performance, how can you do the same within a pipeline?

In Viya you can choose any of the model nodes and download score code for that model. In the Pipeline Comparison you can select a different Champion Model by right clicking on your model and selecting Set as champion. You can also download the score code from here as well.

Recommended Resources

SAS® Enterprise Miner™ and SAS® Visual Data Mining and Machine Learning Hand Shake

SAS9.4 and SAS Viya Functional Comparison

Make SAS® Enterprise Miner™ Play Nicely with SAS® Viya®

Monthly SAS Viya Release Highlights Show

Want more tips? Be sure to subscribe to the Ask the Expert board to receive follow up Q&A, slides and recordings from other SAS Ask the Expert webinars.