Model Studio for SAS Enterprise Miner Users: part 7, Backing up your work

[#1]

This post will discuss the differences and similarities in how to back up your work in SAS Enterprise Miner versus Model Studio. Alright, your move to Model Studio is well underway. You’re attacking the data science lifecycle with a powerful, new interface. A veritable plethora of data preprocessing methods and machine learning models are in your arsenal, and you build models against huge data at lightning quick speed thanks to the CAS engine in SAS Viya. Then your IT department knocks on your door (or more likely sends you an email) and reminds you that you need to back up your work. You know of a few ways to do this for Enterprise Miner, but what about Model Studio? What do you know about Model Studio up to this point in your transition journey? Model Studio has a different interface, a different engine under the hood, it is built on more modern technology. Uh oh! Will this be difficult?

Well, for this one, I have some good news. You can report back to your IT department that you should have your Model Studio work backed up by the end of the day. Although some of the small details differ, the ways to back up work in Model Studio compared to Enterprise Miner, are more similar than they are different.

This post is part 7 in a series I’ve been working on to introduce Model Studio to the SAS Enterprise Miner user. If you haven’t seen the others, you may want to stop and check them out first. Links are at the bottom of this post.

Enterprise Miner:

1. Saving a Diagram

In Enterprise Miner, one way to back up work takes place at the diagram level, where you export, or save, a diagram as an XML file. From within the project window in the upper left corner of an opened project, right-click on the name of the diagram that you want to save. From the pop-up menu, select Save As…

Select any image to see a larger version.
Mobile users: To view the images, select the "Full" version at the bottom of the page.

Then simply enter a file name and select a location where you want the diagram, in the form of an XML file, to be saved.

All of the nodes within the diagram and their property settings are saved in the XML file. To reuse the diagram, simply import it into an existing project from within the project window.

2. Saving a Project

A second way to back up work in Enterprise Miner takes place at the project level. As stated in an earlier post, when a project is created in Enterprise Miner, a specific project folder is created in a location chosen by the user. That project folder contains several subfolders. Contained within the subfolders are all the information about the project, including project and column metadata, details on all diagrams, and even the data the project uses. Because the project folder contains data, it can be very large. However, by saving the folder to a secondary location all the work within the project would be backed up.

The screen capture below shows the project folder for a project called Blog. It is located on the computer’s D drive in a folder called EM Projects. (There is a second project folder in the same location.)

This project folder can then be copied and placed into a second location for back up. The saved project folder can even be brought back into Enterprise Miner as a new project. To do this, you start in Enterprise Miner as if creating a new project. Use the name of the saved project folder and point to the location where the project folder exists. This duplicates the saved project as a new project.

Model Studio:

Like SAS Enterprise Miner, there are two ways to back up work in Model Studio. And although some details are different, the methods in Model Studio are very similar to what is done for Enterprise Miner.

1. Saving a Pipeline

Like saving a diagram in Enterprise Miner, a pipeline can be saved in Model Studio. All nodes and properties are saved when a pipeline is saved. However, Model Studio streamlines this process and makes it easier for the analyst by preconfiguring a specific location where pipelines are saved and shared. This location is known as the Exchange. To save a pipeline to the Exchange, select the Options button (three vertical dots) next to the name of the pipeline to be saved. Then select Save to The Exchange…

You are prompted to provide a name (a required field) and a description (an optional field). You do not need to choose a location.

Once saved, the pipeline can be selected for use from the Exchange. The Exchange is accessed by using the middle short-cut button in the column on the extreme left of the Model Studio interface.

In addition to backing up your work, this is also a great way to share best practices. Compared to Enterprise Miner, Model Studio is designed to be able to share work much easier with colleagues. Suppose you’ve spent weeks working on a pipeline that focuses on data preprocessing. You feel that not only might you use this pipeline in a future project, but that other data scientists on your team could benefit from your work as well. By saving your pipeline to the Exchange it places your custom pipeline in a central repository for others to take advantage of. In fact, although not emphasized here, you can even save a single node to the exchange. If you alter a node by changing property settings, these settings are saved when you save the node to the Exchange.

2. Saving a Project

Again, like Enterprise Miner, a project itself can be exported and saved in Model Studio. To save a project, from the Projects view of Model Studio, click the Options short cut button (three vertical dots) contained within the project tile of the project to save. Then select Export.

Your computer system will indicate that a zipped file is being prepared for download and will be saved in the download folder once complete. The name of the zip file will match the name of the project.

This zip file can then be placed in a secure location to back up the work it contains. Unlike Enterprise Miner, this project zip file does not contain the project data. A bit more on this in a moment.

Model Studio also makes it very easy to import a saved project. Simply click the Additional actions for projects button (three vertical dots) in the upper right corner of the Projects view. Select Import and then the type of Model Studio project you are importing.

The Import window opens. Simply Browse to the project zip file and assign the Data set to be used with the project and Import.

Note the field to select data. Unlike Enterprise Miner where the data is saved within the project folder, the zip file of the saved Model Studio project does not include the data. This makes sense given the size of the in-memory, distributed data that SAS Viya may be using for the project. If the data is so large that it needs to be saved across many worker nodes in CAS memory, there’s no way the same data could be saved within a zipped folder.

Now you know you can tell your IT department not to fret! You’ll have your Model Studio work backed up by the end of the day…maybe even sooner than that. The process of saving a diagram or a project in Enterprise Miner is very similar to saving a pipeline or project in Model Studio. It is not surprising that Model Studio has taken these capabilities and expanded upon them. Model Studio is designed to be able to share best practices, and the Exchange really streamlines this process for pipelines. Even the process of exporting projects as zip files direct from the interface makes backing up projects simple.

Prior Posts:

Model Studio for SAS Enterprise Miner Users: Part 1

Model Studio for SAS Enterprise Miner Users: Part 2, Data

Model Studio for SAS Enterprise Miner Users: Part 3, Let’s get philosophical

Model Studio for SAS Enterprise Miner Users: Part 4, Partitioning Data

Model Studio for SAS Enterprise Miner Users: Part 5, Building Models…Let’s get physical!

Model Studio for SAS Enterprise Miner Users: Part 6, The Joy of Model Comparison

Find more articles from SAS Global Enablement and Learning here.