#imputation

Question 1: Single choice

A house-price table leaves GarageArea blank exactly for the houses that have no garage. Why is filling these blanks with the column mean a poor choice?

It is fine: mean imputation is unbiased whenever fewer than half of the values are missing1
They are not missing at random: the true value is 0, which the mean misstates2
Mean imputation shifts the column's mean, which biases every other house's prediction3
pandas cannot compute a mean for a column that still contains blank values4

Timed test