I have some NaN values in my data and I cannot replace it with median or something else.
Now I have to check out the correlation, using scatter plot and Pearson correlation.
Should I drop these NaNs or there is no difference? Actually, I see the difference, but a very, very small difference. What is the best solution here?