問題文
In a credit dataset, the field for previous defaults is empty for applicants who have no credit history at all. An analyst is deciding how to handle it. Which handling preserves the meaning of the absence?
選択肢
- Fill the field with zero, and an applicant with no credit history has by definition never defaulted and zero is the correct value.
- Add an indicator column marking that the value was absent, in addition to whatever filling is applied to the original column.
- Fill the field with the column mean so the distribution is unchanged, and the absence leaves no trace.
- Drop every row where the field is empty, as rows without a credit history contribute nothing to a default model.