Databricks Machine Learning Associate Question 524
Single answerYou are working with a dataset in Databricks that contains a column for customer age, but many values are missing from this column. The missing values are randomly distributed, and you decide to fill them using an imputation strategy. Which of the following scenarios is MOST appropriate for imputing missing values in this context?
- A
Impute the missing values with the mean age if the distribution of ages is approximately normal.
- B
Impute the missing values with the mode age if the distribution of ages is heavily skewed.
- C
Impute the missing values with the median age if the distribution contains significant outliers.
- D
Impute the missing values with a fixed value, such as 0, to maintain consistency in the dataset.
Show answer and explanation
Correct answer: A
Explanation
When the distribution of ages is approximately normal, the mean is the most accurate and unbiased measure for imputing missing values. Using the mean ensures that the missing values are replaced in a way that preserves the overall statistical properties of the dataset. While other methods may work in specific cases, the mean is generally preferable when dealing with normal distributions.
- A. Correct.
Imputing with the mean is appropriate for a normal distribution because it minimizes the difference between actual values and the imputed values, making it the most accurate in this case.
- B. Incorrect.
Imputing with the mode is not suitable for numerical data like age, as the mode represents the most frequent value and does not account for overall distribution trends.
- C. Incorrect.
Imputing with the median is better for distributions with outliers, but the question specifies no mention of outliers, so this is not the most appropriate option.
- D. Incorrect.
Imputing with a fixed value such as 0 can introduce bias and distort the distribution, making it inappropriate for this scenario.