Should I normalize my nan_mask or leave them as 1's and 0's for the model?

Question

I have a ML project in which some data rows are missing data. In this case, I replace the missing data with the average of the column, add a nan_mask column for each column