Changing mutual information score with shuffling of rows

I am working on a dataframe to conclude something and i have around 2400 columns and i want to reduce that number and for that i am calculating mutual information score and after that calculating VIF score.
Though mutual information score is changing when i am shuffling the dataframe rows but how is that possible because the values in all columns are still the same and the relative ordering among columns is not changing then why after shuffling the rows the mathematical calculations are somehow changing and i am getting a different mutual information score every time i run the cod.

Hi Utsav_Sharma1,

My guess would be that this results from rounding differences due to different sequences of calculations.

I have the same problem
How did you perform the code?