Notes on Correspondence Analysis of Power Transformed Data Sets

TL;DR

This paper explores the application of correspondence analysis to power and Box-Cox transformed data, focusing on datasets with zeros to clarify the method's interpretation and effectiveness.

Contribution

It provides a detailed examination of correspondence analysis for datasets with zeros after power transformations, an area less addressed in prior research.

Findings

Clarifies the interpretation of CA on zero-including transformed data

Distinguishes between positive and zero-including datasets in CA analysis

Offers insights into the visualization and analysis of transformed data sets

Abstract

We prospect for a clear simple picture on CA of power transformed or the Box-Cox transformed data initiated since 2009 by Grenacre. We distinguish two types of data sets : strictly positive and with zeros; we concentrate on the latter.