I want to split the data into train and test when i first encounter the NaN value in a particular column how to do that ?