How to use K-NN imputer without replacing with decimal values example ( 0.75,0.6) instead of binary outcome (yes or no, 1 or 0)?

I am trying to impute some missing categorical values using K-NN imputer, after imputation the missing values are replaced with some decimal numbers. I want to use K-NN as classifier and the output (imputation) that I want was only 2 outcome (0,1). Does anyone know how to perform K-NN imputation as classifier using scikit learn library I am newbie to data science and have already read the documentation but no help.