×

Warning

The RokSprocket Module needs the RokSprocket Component enabled.

Partial Least Squares Regression (PLS)

Partial Least Squares Regression (PLS)

Classical linear regression is inappropriate when the predictors are correlated (multicollinearity). Learn the principe of PCR and PLS regression designed to deal with multicollinearity and when it is relevant to use them.

Course Details

Learning Objectives

Learning Objectives:

Upon completion of this module, participants will be able to: [list bullet-5]

  • Understand the context of use for PCR and PLS regression
  • Understand what the underlying assumptions of the techniques are
  • Construct the regression models
  • Assess the goodness-of-fit of the model to the data
  • Identify common issues for these regression models, diagnose problems and fix them
  • Interpret statistical software output
  • Target Audience

    Target Audience:

    This module is aimed at all scientific staff who collect data and who must make decisions based on them. The regression techniques covered in this session will be particularly useful for people who are interested in relating/predicting a variable to/from a set of correlated explanatory variables.

    Prerequisite

    Prerequisite:
    • Participants must know the mechanism underlying simple and multiple linear regression techniques, interpretation of the data, model validation, common problems in regression, and so on. Either by having attended the training session Linear and Multiple Linear Regression Techniques or by possessing a similar background.
    • Participants must also know the mechanism underlying principal component analysis as this multivariate technique is used in both advanced methods covered in this session. Either by having attended the training session Principal Component Analysis and its Applications or by possessing a similar background.

    Course Outline

    Course Outline:

    The Issue of Multicollinearity in Regression Models

    • What is Multicollinearity?
    • Consequence on Multiple Linear Model
    • Diagnostic Tools
    • Alternative Tools: PCR and PLS
    • Regression on Principal Components (PCR)
      • Context of Use and Principle
      • The PCR Essentials
      • PCR in Practice
      • Selection of a Subset of the Principal Components
      • Use and Interpretation of PCR Results
      • Software Packages for PCR
    • «Partial Least Squares» PLS Regression
      • Context of Use and Principle
      • Applications
      • Working with Several Response Variables?
      • Steps involved in the method
      • Extraction of principal components
      • Use and Interpretation of PCR Results
      • Software Packages for PCR
    • Pratical Implementation of PLS
      • Initial Data
      • General Principle
      • Selection of principal components
      • Type of results generated by statistical software
      • Illustrations with Case Studies
    • Conclusion
      • PLS : An alternative to multiple linear regression
      • Scope of PLS regression
      • Limitations of PLS
      • Some references

    Practical Info

    Practical Info:

    Recommended Duration: 1 day

    Related Sessions

    • An applied set of modules with focus on the most widely used multivariate methods and their applications in several fields of application. Learn about the principle of the methods, the data needed, and the information they provide.

    • Learn about preference mapping techniques to explore and understand consumer preferences. Applications dealing with segmentation and the identification of niche markets are discussed. Focus on pitfalls and good practices.

    • Predictive analytics (PA) is on everyone's lips. But what is it really all about? Discover its principle, implementation, typical pitfalls and good practices. Learn about data wrangling and munging, a crucial step in predictive analytics. An overview of the most commonly used models is also presented.

    • The primary goal of this method is to discover which variables have the best ability of discriminating between two or more known groups in your data. Discrimimant analysis may also be used to build predictive analytics models.