I agree with dr. Zuur in as far as applying DFA to binary data-not re

I agree with dr. Zuur in as far as applying DFA to binary data-not reccomended 
is concerned and I am aware that other methods such as kernel DA or logistic 
regression may be more appropriate. Klecka (1980) and Hand (1983) pointed out 
that violating homogeneity assumtion is however not likely to affect the 
conclusions based on DFA. The purpose of the analysis I run was basically 
to cluster plants groups and associate them to already established lake groups 
and I am not quite familiar with the other methods. Are they able to do this?
 
My data also covers other variables sets such as invertebrates richnes (with 
homogeneity assumtion not violated) on which I separately run DFA. 
However, I still have unclear the interpretation of  classification function 
coefficients (or Fisher linear discriminant function). They represent the 
likelyhood that an object (variable/ invertebrate taxa) pertains to a 
particular class. Is in this case coefficient 0.1 (and e.g. -2) biger that -25?
 
Thanks

Dragos
  

Dragos Zaharescu
Vigo University

Hand DJ. 1983. A Comparison of Two Methods of Discriminant Analysis Applied to 
Binary Data. Biometrics. 39 (3), 683-694
 Klecka, William R. (1980). Discriminant analysis. Quantitative Applications in 
the Social Sciences Series, No. 19. Thousand Oaks, CA: Sage Publications.




________________________________
From: Highland Statistics Ltd. <[email protected]>
To: [email protected]
Sent: Wednesday, August 5, 2009 9:35:06 AM
Subject: Re: [ECOLOG-L] Interpretation of Fisher's classification function 
coefficients of discriminant analysis

On Tue, 4 Aug 2009 09:46:17 -0700, Dragos Zaharescu <[email protected]>
wrote:

>I run discriminant analysis on a dataset with plants speci
>Hi all,

>I run discriminant analysis on a dataset with plants species presence as 
>variables (binary, independent) and lakes # as cases. I also have a
>classification variable with 4 classes. 




Assuming you apply "ordinary" discriminant analysis, you are violating the
homogeneity assumption if you apply DA on binary data. See Chapter 14 in
Zuur et al (2007) "Analysing Ecological Data", for all 10-ish assumptions of DA.





>I have some problems in interpreting Fisher's classification function
>coefficients. 
>It is said that a case belongs to the class for which the score is highest. 


No..that is wrong as well...see the same chapter. It is about the distance
of a point to the centroids.


Alain



Dr. Alain F. Zuur
First author of:

1. Analysing Ecological Data (2007).
Zuur, AF, Ieno, EN and Smith, GM. Springer. 680 p.
URL: www.springer.com/0-387-45967-7


2. Mixed effects models and extensions in ecology with R. (2009).
Zuur, AF, Ieno, EN, Walker, N, Saveliev, AA, and Smith, GM. Springer.
http://www.springer.com/life+sci/ecology/book/978-0-387-87457-9


3. A Beginner's Guide to R (2009).
Zuur, AF, Ieno, EN, Meesters, EHWG. Springer
http://www.springer.com/statistics/computational/book/978-0-387-93836-3




Reply via email to