Tree-Based Machine Learning Methods in SAS® Viya®

Predictive modelers and data analysts who want to build decision trees and ensembles of decision trees using SAS Visual Data Mining and Machine Learning in SAS Viya

• Build tree-structured models, including classification trees and regression trees
• Use the methodology for growing, pruning, and assessing decision trees
• Build tree-based ensemble models, including forest and gradient boosting
• Run isolation forest and Poisson and Tweedie gradient boosted regression tree mogels
• Provide an introduction to deep forest models
• Implement open source in SAS and SAS in open source
• Use decision trees for exploratory data analysis, dimension reduction, and missing value imputation

Before attending this course, you should have the following:

• An understanding of basic statistical concepts. You can gain this knowledge from the SAS® Visual Statistics in SAS® Viya®: Interactive Model Building course
• Familiarity with SAS Visual Data Mining and Machine Learning software. You can gain this knowledge from the Machine Learning Using SAS Viya course

Introduction to Decision Trees

• Tree-structured models
• Tree-based models in SAS Viya
• Regression trees

Growing a Decision Tree

• Recursive partitioning
• Split search
• Splitting criteria
• Missing values and variable importance

Preventing Overfitting in Decision Trees

• Pruning
• Cross validation
• Autotuning
• Secondary uses of decision trees

Ensembles of Trees: Bagging, Boosting, and Forest

• Bagging and boosting
• Forest models
• Open-source random forest models
• Isolation forest models
• Introducing deep forest models

Tree-Based Gradient Boosting Machines

• Overview of gradient boosting
• Tuning a gradient boosting model
• Gradient boosting for transfer learning
• Gradient boosted Poisson and Tweedie regression trees
• SAS gradient boosting using open source

A Practice Case Study

• Data exploration
• Creating decision trees
• Out-of-time testing
• Creating tree-based ensemble models
• Hyperparameter tuning
• Model comparison