Question - You are given a data set consisting of variables with more than 30 per cent missing values. How will you deal with them?
Answer -
If 30 per cent data is missing from a single column then, in general, we remove the column. If the column is too important to be removed we may impute values. For imputation, several methods can be used and for each method of imputation, we need to evaluate the model. We should stick with one that model which gives us the best results and generalises well to unseen data.