Dear All,
I would like to ask some help. Something is not clear for me during the v-fold crossvalidation in the tree methods (I would like to use CRT and CHAID). For me it seems that the v-fold crossvalidation method in SPSS decision trees is different from the v-fold crossvalidation decribed in Breiman et al. (1998). According to Breiman et al. (1998) the model is developed v times only on the (v-1)/v of the total dataset and validated on the left out subsample. Therefore v (more or less) different models are given at the end of the process. If SPSS does the same v-fold crossvalidation as it is described above, could someone please write it to me or suggest some reference on how the single model is given at the end of the process?
Kind greetings,
Brigitta
Tóth Brigitta
Tel: +36 30 322 80 54
+39 347 22 69 181
E-mail: [hidden email]