EDA (NEDA) is done through the NBO program. If I'm not mistaken NBO version3 (which is built into g09) does not support it. NBO version 6 does support it and g09 does have interface to it through population keyword, but you need the NBO6 software itself (not free).
It is not going to be very easy to do it in G09 using NBO. You can switch to some other free codes like Gamess or PSI4. But I did not understand the part of your question quoted "based on the atom fragmentation".
Can we compare the results obtained by SAPT(0) by Psi4 and EDA-NOCV by ADF?
I mean, I want to compare the energy values (Electrostatic, Dispersion) obtained by EDA-NOCV method with SAPT method. I do not have ADF, but I have Psi4.