GA for feature selection get different results

inceptorfull

Hi all,

I have dataset contains 37 attributes
I used neural network and logisitc regression on the FULL DATASET and got their accuracy rate

now I want to use GA to select the best attributes so I can RE-apply NN and logisitc regression and get accuracy rate ( suppose to be higher )

anyway ,
when you the optimization selection Evolutionary,  it requires model to be inserted , I used NN and got higher accuracy but when used such inputs again it gives me different accuracy and same for logisitic regression

so what model should I use to train the GA so I can take its output to be used as inputs for both NN and logisitc?