2

How can I run simple correspondence analysis (CA) in Python? In the sklearn library, there only appears to be multiple correspondence analysis (MCA) and canonical correspondence analysis (CCA) options. However, my data is not categorical and does not need the additional linearity constraints applied by CCA.

My impression based on this link on CCA and this one on MCA is that regular CA cannot be applied by using one of the two other option.

FFT
  • 929
  • 8
  • 17
  • 2
    A tutotrial: [tutorial to build a CA class](http://okomestudio.net/biboroku/?p=2357) And the canonical CA: [see also here](http://scikit-learn.org/stable/modules/generated/sklearn.cross_decomposition.CCA.html) – seralouk Jul 11 '17 at 07:33
  • Thanks, the tutorial looks great! On the other hand, is it possible to use the sklearn canonical CA for regular CA? – FFT Jul 11 '17 at 23:12
  • 1
    I will write an answer adding more resources ! – seralouk Jul 12 '17 at 05:56
  • 1
    Note that the CCA method in scikit-learn stands for canonical correlation analysis, which slightly differs from canonical correspondence analysis, see for a more detailed explanation [here](https://stats.stackexchange.com/questions/38629/difference-between-canonical-correpondence-analysis-and-canonical-correlation-an). – Archie Feb 18 '20 at 10:54

1 Answers1

1

For correspondence analysis you can use a package "prince" https://pypi.org/project/prince/ . It makes a plot from the data in a few steps. Although the drawback is that you (I) cannot see details of the calculated residuals for instance.

Leo
  • 199
  • 2
  • 4