I have an input data as follows:
coordinates = [0, 4, 0, 5, 4.5, 3, 6, 4.5, 6, 2, 6, 1, 0, 1, 0, 2]
Demand = D1 D2 D3 D4
P1 0 20 0 30
P2 0 0 20 0
P3 0 0 0 10
P4 30 0 0 0

when I train the NN using only the coordinates it does well but when i include the demand the training process becomes poor. I also normalized everthing but it did not work. The demand is unchangeable in episode level so it harming the learning process. Do there is a way to concatenate the two data in a way it does not harm the learning process??