Normalizing the dataset results in a single class prediction #17

I have used your implementation on my own dataset. It seems to work fine when the data is in its original format. However, whenever I normalize or standardize the data, it always predicts only one class (0 in my case), i.e., all the predictions are 0. I cannot seem to understand why is that. Do you have any idea?