WEKA training and cross validation

I am trying to use the python weka wrapper. I am using the Cross validation method. It then prints the classification results. Then i use build_Classifier and test on training data using test_model. It provides different no.of classification instances compared to the cross validation model.

From what i understood, in the cross validation model, 10 different models are built, and then the accuracy is averaged while the models are discarded. Then it fits the entire data again and produces the classification results.

then when the data is the same, shouldnt i get the same results with the build_classifier model as well? or is it because i put randomstate in crossvalidation but did not randomize the data in build_model?

from Recent Questions - Stack Overflow https://ift.tt/2XFHW69
https://ift.tt/eA8V8J