Back to photostream

Correspondence Analysis (CA) Plots

Principle of CA is to synthesize information contained in a contingency table (C columns x L rows), by focusing on relative values rather than on raw values, i.e. correspondences between modalities of 2 variables.

 

CA allows to detect axes by which points (= modalities of qualitative variables) spread the most from the center of inertia. These "synthetic" axes may result of influence of one or several variables or observations.

 

Thus, by projection of points (observations or variables) on plans defined by these new axes, we have a representative "photograph" of points.

 

By analogy, imagine someone is trying you to recognize an animal drawn on a paper sheet (that is a camel) but you see only its face shape. You can not conclude if it is a camel or a dromedary, just because angle of analysis is not the most informative. You need a profile image of the animal to conclude with certitude that it is a camel. On the same way, CA calculates new axes (and then plans) able to advise you at the best about repartition of points.

 

To download a free demo of the StatEL software, please visit www.adscience.eu

474 views
0 faves
0 comments
Uploaded on December 16, 2008
Taken on December 6, 2008