Discover SAS Visual Data Mining and Machine Learning Procedures

[#1]

Welcome to you, early explorer for SAS Visual Data Mining and Machine Learning! This offering includes data mining and machine learning capabilities that run on the SAS Viya platform in an in-memory distributed computing infrastructure called Cloud Analytic Services—CAS. SAS Visual Data Mining and Machine Learning provides a single environment for data scientists to perform tasks associated with data preparation, feature engineering (variable manipulation and selection), model training, assessment, and deployment.

Your interface into this product is SAS Studio, the web-based user interface for SAS.  You can perform your statistics and machine learning work by clicking through prompted SAS Studio Tasks, or by writing SAS code.

Before we continue – are you new to machine learning? If so, check out this resource: What is machine learning and why does it matter?

This introductory post provides a sort of table-of-contents to the SAS Visual Data Mining and Machine Learning procedures that are available today. Look for more articles coming soon, with more details about each task and how you might want to use them all together.

I’ve divided these procedures into four major categories: Data Wrangling, Unsupervised Learning, Supervised Learning, and Model Assessment and Scoring.

Data Wrangling

Here are the data wrangling procedures, including the SAS Studio Task where you will find each one (in the “Prepare and Explore” category) and a link to the SAS documentation.

• Variable binning (PROC BINNING, Task: Binning)
• Variable cardinality analysis (PROC CARDINALITY, Task: Summary)
• Sampling and partitioning (PROC PARTITION, Tasks: Partition, Summary)
• Missing value imputation (PROC VARIMPUTE, Task: Imputation)
• Variable selection (PROC VARREDUCE, Task: Variable Selection)
  
  

Unsupervised Learning

Here are the unsupervised learning procedures, including the SAS Studio Task where you will find each one (in the “Unsupervised Learning” category) and a link to the SAS documentation.

• k-means & k-modes clustering (PROC KCLUS, Task: Clustering)
• Principal component analysis
• Standard PCA  (PROC PCA, Task: Principal Component Analysis)
• Robust PCA (PROC RPCA, Task: Robust Principal Component Analysis)
• Moving Window PCA (PROC MWPCA, Task: Moving Window Principal Component Analysis)
• Support Vector Data Description (PROC SVDD, Task: Support Vector Data Description)
• Text Mining (PROC TEXTMINE, Task: Text Parsing and Topic Discovery)
  
  

Supervised Learning

Here are the supervised learning procedures, including the SAS Studio Task where you will find each one (in the “Supervised Learning” category) and a link to the SAS documentation.  Supervised learning is a big category! I’ve divided this list into subcategories to help you navigate (they are not categorized this way in SAS Studio).

Regressions

• Ordinary least squares models (PROC REGSELECT, Task: Linear Regression)
• Logistic regression models (PROC LOGSELECT, Task: Logistic Regression)
• Generalized linear models (PROC GENSELECT, Task: Generalized Linear Models)
• Partial least squares regression models (PROC PLSMOD, Task: Partial Least Squares Regression)
• Quantile regression models (PROC QTRSELECT, Task: Quantile Regression)
• Nonlinear regression models (PROC NLMOD, Task: N/A)

Neural Networks

• Neural network models (PROC NNET, Task: Neural Network)

Tree-related

• Decision tree models (PROC TREESPLIT, Task: Decision Tree) – single tree
• Forest models (PROC FOREST, Task: Forest) – ensemble method
• Gradient boosting models (PROC GRADBOOST, Task: Gradient Boosting) – ensemble method

Other

• Support vector machine models (PROC SVMACHINE, Task: Support Vector Machine)
• Factorization machine models (PROC FACTMAC, Task: Factorization Machine)
• Boolean rule model (PROC BOOLRULE, Task: Boolean Rules)
  
  

Model Assessment and Scoring

• Assess supervised models (PROC ASSESS)
• Scoring models (Data step, Task: Scoring; PROC ASTORE, Task: N/A)
• Scoring of new text docs (PROC TMSCORE, Task: N/A)
  
  

I hope this gives you a good idea of the initial inventory of methods with SAS Visual Data Mining and Machine Learning. In future posts, I’ll go into more detail about how each method is used.

[Stewartli]

Hi Beth, 

When or how can we use VDMML via SAS Academic on demand? Is there any other way I can access Auto-tuning function? 

Thank you. 

Stewart