Hierarchical and Optimized Models
Hierarchical models and optimized models represent two approaches to improve model performance without invoking non-linear methods. Because models work best when given simple tasks, hierarchical and optimized models are used when standard models perform poorly on data from complicated systems. Hierarchical models utilize empirical or first-principle-based segregation of the data to break the problem into smaller, easier to model portions. They are frequently used with both classification and regression problems. Optimized models use combinatorial methods to identify preprocessing and modeling conditions that provide the most flexible and stable models for a given task. This course will discuss concepts and practical implementation of both methods to improving model performance. The course includes hands-on computer time using PLS_Toolbox or Solo for participants to understand better the differences between the various options.
1.0 Introduction to Hierarchical and Optimized Models
2.0 Hierarchical Modeling for Classification
3.0 Hierarchical Modeling for Regression
4.0 Model Optimization
5.0 Evaluating Optimized Models