You can only validate using data that was not included in the
original classification.  However, you might consider using the
"Jackknife" approach. You reserve one (or a few) cases and run the
classification without them, then identify them for validation.  You
can do this for all the cases.  Check to make sure that the
classification functions are not changing significantly with each
run.  Many classification data sets have lots of redundancy, which is
needed for this approach.

Jim Palmer, SUNY ESF, Syracuse, NY

>Hello,
>
>I recently read that:
>you can't validate the "classification model with the data used to develop
>the model. You must use completely independent data otherwise you bias the
>results.
>
>Is there any resampling approach to address this issue?
>I would be grateful if any of you can point me to some good references or
>studies.
>
>Thanks for your help
>
>Henry


--
\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\

                      James F. Palmer, Professor
                  Faculty of Landscape Architecture
          SUNY College of Environmental Science and Forestry
                         Syracuse, NY  13210
   voice: 315 470-6548            internet: [log in to unmask]

\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\%\