1
$\begingroup$

I was studying by my own correspondence analysis and I got some questions about the map that one gets using this method for some rows and columns. For example the following map: http://www.statmethods.net/advstats/images/ca3.jpg

The questions that are troubling me are the following:

  1. The quadrants have relevant importance? I mean, that some words are in one quadrant and not in another. I thought that the most important thing in correspondence analysis were distances between the words and the origin.
  2. The distances between words are important, true? Not only between one word and the origin.
  3. Is the same to put the words in the symmetric way? I mean, "Brown" and "Black" to be in the upper quadrant and "Red", "Haze," and "Brown" downstairs, and the same with the righthandside. With this I wanted to know if is relevant or not to be in a negative quadrant.

I will be really grateful if someone can explain me this doubts, because I have spend many days reading articles and books and I don't get this information. I'm actually interested in buying the book "Correspondece analysis in practice", by Michael Greenacre.

$\endgroup$
0
$\begingroup$

Do not worry about the quadrants indeed, orientation in biplots is mostly arbitrary. Variables can be in different quadrants and still close together, or in the same quadrant but far apart. The distance between the variables is indeed of interest as well, it indicates e.e how different people with brown and black hair are with respect tot eye colour. The distance to the origin relfects how different they are from the (weighted) average eye colour profile.

I can really recommend Greenacre's work, take a look at this online available book too.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.