HomeScience & TechnologyRelated VideosMore From: SAS Software

SAS Enterprise Miner: Impute, Transform, Regression & Neural Models

76 ratings | 43640 views Chip Robie of SAS presents the fourth in a series of six "Getting Started with SAS Enterprise Miner 13.2" videos. This fourth video demonstrates imputing and transforming data, building a neural network, and building a regression model with SAS Enterprise Miner. For more information regarding SAS Enterprise Miner, please visit SAS ENTERPRISE MINER SAS Enterprise Miner streamlines the data mining process so you can create accurate predictive and descriptive analytical models using vast amounts of data. Our customers use this software to detect fraud, minimize risk, anticipate resource demands, reduce asset downtime, increase response rates for marketing campaigns and curb customer attrition. LEARN MORE ABOUT SAS ENTERPRISE MINER SUBSCRIBE TO THE SAS SOFTWARE YOUTUBE CHANNEL ABOUT SAS SAS is the leader in business analytics software and services, and the largest independent vendor in the business intelligence market. Through innovative solutions, SAS helps customers at more than 75,000 sites improve performance and deliver value by making better decisions faster. Since 1976 SAS has been giving customers around the world The Power to Know.® VISIT SAS CONNECT WITH SAS SAS ► SAS Customer Support ► SAS Communities ► Facebook ► Twitter ► LinkedIn ► Google+ ► Blogs ► RSS ►
Get embed code!
Text Comments (12)
Michael Tuchman (8 months ago)
How does Impute compare with PROC MI in SAS/STAT?
SAS Software (8 months ago)
That’s a broad question that can be answered in many ways! We'll keep things basic and let you decide how deep you want to go. SAS Enterprise Miner’s Impute node uses a combination of SAS procedure and function calls to perform imputations for missing variable values in a data set. The SAS code is behind the scenes, embedded within configurable GUI objects that you set up to perform specified Impute operations. You do not need to know SAS programming details to configure the Enterprise Miner Impute node. Here is a link to the SAS Help Center overview of the SAS Enterprise Miner Impute node: The SAS/STAT MI procedure is a licensed component of the SAS/STAT product and the SAS programming language. Like the Impute node in Enterprise Miner, you use PROC MI to specify an imputation strategy for missing variable values. Unlike Enterprise Miner, SAS/STAT is not a GUI application, it is a set of analytic programming tools. You need to understand SAS programming and SAS syntax to effectively use SAS/STAT and PROC MI. Here is a link to the SAS Help Center overview of PROC MI for SAS/STAT: Comparing the SAS documentation overviews of SAS/STAT PROC MI and SAS Enterprise Miner’s Impute node is the best way for you to determine the fundamental and functional differences between the two data mining tools to your own degree of satisfaction. If you have any further, more specific questions, please emails us at or feel free to post to the Communities where SAS experts are on hand to help with programming/content related questions.
Michael Tuchman (8 months ago)
This answers my question "why do we impute after partitioning, but not before".
dilnoza yusupova (1 year ago)
Thank you for the video! Do you have a video about anomaly detection model?
SAS Software (1 year ago)
Dilnoza, thank you for your inquiry! We do not have a video on this topic but this white paper may help
Vangelis Dalucas (2 years ago)
Thank you for this video! I usually hear about imputation as a step in predictive modelling using trees/regression/neural networks but what about exploration through clustering? Should we impute before clustering? If yes, above which missing percentage should we consider imputing? Thanks
SAS Software (2 years ago)
Absolutely, you are most welcome! Definitely utilize the SAS Communities for programming / content related inquiries, we're here to help!
Vangelis Dalucas (2 years ago)
Thank you for this reply, well I could feel comfortable making certain assumptions as my data set has more than 160000 observations. That made me think that even if I have 90% missing percentage I could still impute (using mean values) and not reject. But doing nothing, either Imputing or rejecting, does not seem right...
SAS Software (2 years ago)
That’s a good question! It depends on the number of missing observations you have in your data, and the assumptions you feel comfortable making about them. Since the point of clustering is to place objects into groups suggested by the data, the clearer your data is, the better your clustering results might become. If you feel comfortable making certain assumptions about missing observations, imputing your data could improve your clustering analysis. However, as part of your clustering analysis, you might want to know about the missing values, and the missing values themselves might be an important component to your analysis if they are indicative of anything in particular that should be considered. You could try both and see if you get similar results. It’s hard to say whether imputation is appropriate for the circumstances without knowing more about why the data has missing values in the first place. For example, are the missing values the result of a random technical problem, or are they missing for a particular reason that shouldn’t be removed from the analysis… In SAS Enterprise Miner 13.2, there is a “Scoring Imputation Method” property on the Cluster Node, which helps you select how to treat missing observations. See the “Cluster Node Train Properties: Missing Values” property section, and the documentation on the Cluster Node in the SAS Enterprise Miner help for more information on imputation options. Hope this helps! Should you have usage / programming inquiries, you can certainly post to where SAS experts are on hand to assist!
SAS Software (2 years ago)
We're looking into this for you, Vangelis!
Philip Wang (3 years ago)
This is an excellent presentation to learn SAS EM!
SAS Software (3 years ago)
We're certainly glad you are enjoying the content and thank you for your post!