Home / ML Associate practice test / Data Processing / Question

Databricks Machine Learning Associate · free practice question

A data scientist has a loan_amount column with 8% missing values. Exploratory analysis reveals that the distribution is heavily right-skewed due to a small…

Exam-style question on Data Processing from the free ML Associate practice test, with the answer and a full explanation below.

Updated November 10, 2026

Data Processing

A data scientist has a loan_amount column with 8% missing values. Exploratory analysis reveals that the distribution is heavily right-skewed due to a small number of very large loan values. They must choose between mean and median imputation. Which answer correctly explains the trade-off and identifies the more appropriate strategy for this scenario?

  1. AMedian imputation is more appropriate, because the median is a resistant measure that is not influenced by extreme values in the upper tail and better represents the typical loan amount for most borrowers
  2. BMean imputation is more appropriate, because the mean is pulled upward by the long upper tail and reflects the high-value loans, which makes it a more representative central estimate for this population
  3. CEither is equally appropriate, because mean and median produce identical imputed values here: skewness changes the relative frequency of tail values but not the location of the mean or the median
  4. DMode imputation is more appropriate than both, because the mode is the most frequently observed loan value, which makes it the best proxy for a missing amount whatever the shape or skew of the distribution
  5. EMean imputation is more appropriate, because the mean uses the full set of data points while the median uses the rank-ordered middle value, which makes the mean the more efficient and unbiased estimator here
Answer & explanation

Correct answer: A

WHY A: The median is the appropriate choice for right-skewed data because it is a resistant statistic: extreme values in the upper tail do not move it. In a right-skewed distribution the mean is inflated above the typical central value, so median imputation gives a more representative estimate of what a missing loan amount would realistically be.

WHY NOT B: The mean is pulled upward by the right tail, so it overestimates the central value for most observations; replacing missing values with an inflated mean introduces systematic upward bias.

WHY NOT C: This is factually incorrect: in right-skewed distributions the mean is greater than the median, a fundamental property of positive skew rather than an edge case.

WHY NOT D: Mode imputation is primarily suited to categorical variables or discrete distributions with a clearly dominant peak; for continuous right-skewed data the mode often sits at the very low end and is generally not recommended.

WHY NOT E: The mean does use more data points, but it is not universally superior: it is a biased estimator of the typical value in skewed distributions, where the median is specifically preferred for imputation.

Take the full ML Associate practice test free →

Would you pass today?

This question is one of 45 in the free, timed Databricks Machine Learning Associate practice test. Take it and get a section-by-section score.

Start the free test →