spssforum.com

Hi everyone,
I am working on analysis of stable isotope ratios for a population of humpback whales. I have used discriminant function analysis (DFA) on feeding animals and would like to apply the functions to breeding animals to predict feeding location. Does anyone know how to apply previous DFA results/functions to new cases?
Thanks!

Bree

Other than applying the derived model (discriminate equation) to the new cases there isn't one directly, that I can think of. However, there isn't anything wrong with re-running the analysis with the whole dataset then running a split-file and compare the results.

If you find something different do let us know.

_________________
NOTE: Please read the Posting Guidelines and always tell us your OS, the SPSS version and information about your study and data!


Statman
Statistical Services
IBM/SPSS Beta Site

Thanks for the reply statman. Can you elaborate a bit? I am unclear how to re-run the analysis with the whole data set. The whole idea is that I don't know which feeding area the breeding animals belong to but would like to apply the results from the feeding analysis to breeding animals. Does that make sense?

You might be able to set you own code (e.g., = 1) for know and a 2 for unknown then rerun your DA using the same IVs but the DV as the new coded variable. See how the classification table looks and save the set of variables created from DA (group classification & prob, ...). Then do a table or crosstab with the group assignment for DA in the column and various characteristic variable in the rows to see how it compares.

Clear?

_________________
NOTE: Please read the Posting Guidelines and always tell us your OS, the SPSS version and information about your study and data!


Statman
Statistical Services
IBM/SPSS Beta Site