Marie Gaudard A.

Discovering Partial Least Squares with JMP


Скачать книгу

One-Factor NIPALS Model

       Two-Factor NIPALS Model

       Variable Selection

       SIMPLS Fits

       Choosing the Number of Factors

       Cross Validation

       Types of Cross Validation

       A Simulation of K-Fold Cross Validation

       Validation in the PLS Platform

       The NIPALS and SIMPLS Algorithms

       Useful Things to Remember About PLS

       Chapter 5 Predicting Biological Activity

       Background

       The Data

       Data Table Description

       Initial Data Visualization

       A First PLS Model

       Our Plan

       Performing the Analysis

       The Partial Least Squares Report

       The SIMPLS Fit Report

       Other Options

       A Pruned PLS Model

       Model Fit

       Diagnostics

       Performance on Data from Second Study

       Comparing Predicted Values for the Second Study to Actual Values

       Comparing Residuals for Both Studies

       Obtaining Additional Insight

       Conclusion

       Chapter 6 Predicting the Octane Rating of Gasoline

       Background

       The Data

       Data Table Description

       Creating a Test Set Indicator Column

       Viewing the Data

       Octane and the Test Set

       Creating a Stacked Data Table

       Constructing Plots of the Individual Spectra

       Individual Spectra

       Combined Spectra

       A First PLS Model

       Excluding the Test Set

       Fitting the Model

       The Initial Report

       A Second PLS Model

       Fitting the Model

       High-Level Overview

       Diagnostics

       Score Scatterplot Matrices

       Loading Plots

       VIPs

       Model Assessment Using Test Set

       A Pruned Model

       Chapter 7 Equation Chapter 1 Section 1Water Quality in the Savannah River Basin

       Background

       The Data

       Data Table Description

       Initial Data Visualization

       Missing Response Values

       Impute Missing Data

       Distributions

       Transforming AGPT

       Differences by Ecoregion

       Conclusions from Visual Analysis and Implications

       A First PLS Model for the Savannah River Basin

       Our Plan

       Performing the Analysis

       The Partial Least Squares Report

       The NIPALS Fit Report

       Defining a Pruned Model

       A Pruned PLS Model for the Savannah River Basin

       Model Fit

       Diagnostics

       Saving the Prediction Formulas

       Comparing Actual Values to Predicted Values for the Test Set

       A First PLS Model for the Blue Ridge Ecoregion

       Making the Subset

       Reviewing the Data

       Performing the Analysis

       The NIPALS Fit Report

       A Pruned PLS Model for the Blue Ridge Ecoregion

       Model Fit

       Comparing Actual Values to Predicted Values for the Test Set

       Conclusion

       Chapter 8 Baking Bread That People Like

       Background

       The Data

       Data Table Description

       Missing Data Check

       The First Stage Model

       Visual Exploration of Overall Liking and Consumer Xs

       The Plan for the First Stage Model

       Stage One PLS Model

       Stage One Pruned PLS Model

       Stage One MLR Model

       Comparing the Stage One Models

       Visual Exploration of Ys and Xs

       Stage Two PLS Model

       Stage Two MLR Model

       The Combined Model for Overall Liking

       Constructing the Prediction Formula

       Viewing the Profiler

       Conclusion

       Appendix 1: Technical Details

       Ground Rules

       The Singular Value Decomposition of a Matrix

       Definition

       Relationship to Spectral Decomposition

       Other Useful Facts

       Principal Components Regression

       The Idea