Speeding up the search for negative train instances in the data frame

Question

It takes a long time to go through this code. I was wondering if there isn\'t a faster way to iterate through this code. Hence my question, how could you make this code fast