On Components, Latent Variables, PLS and Simple Methods: Reactions to Rigdon's Rethinking of PLS

Area of Science:

• Statistics
• Data Analysis
• Methodology Development

Background:

• Partial Least Squares (PLS) is a statistical method with known limitations.
• Component-type methods have historical challenges in data analysis.
• Previous research suggests modifying or replacing existing PLS methods.

Purpose of the Study:

• To explore the implications of transforming Partial Least Squares (PLS) into a component-based methodology.
• To review historical issues associated with component-type methods.
• To identify and suggest improved methodologies for PLS.

Main Methods:

• Historical analysis of component-type methods.
• Conceptual development of implications from Rigdon's (2012) suggestion.
• Evaluation of alternative statistical models and methods.

Main Results:

• Rigdon's (2012) suggestion to 'kill' PLS and create a component-based method has significant implications.
• Historical problems with component-type methods are documented.
• Maintaining and improving PLS is viable, alongside using alternative methods when appropriate.

Conclusions:

• It is beneficial to improve Partial Least Squares (PLS) while also embracing alternative methods.
• Huang's (2013) PLSe2 methodology is a promising candidate for enhancing PLS.
• The choice of method should depend on the specific data analytic situation.