i'm trying to use anchortabular.
i made the prediction model and a few columns have missing values. but i realized, the anchor rule includes NaN values if the column has more than 1 missing observation. In this case, do i need to do imputation before modeling? but the thing is, my columns that have missing values are meaningful with the missing value itself. Then may be imputation as sentinel value?
i'm trying to use anchortabular.
i made the prediction model and a few columns have missing values. but i realized, the anchor rule includes NaN values if the column has more than 1 missing observation. In this case, do i need to do imputation before modeling? but the thing is, my columns that have missing values are meaningful with the missing value itself. Then may be imputation as sentinel value?