Databricks Machine Learning Associate Question 518
Select 2You are working on a machine learning project with a dataset that contains missing values in a numeric column. You are deciding whether to impute the missing values using the mean, median, or mode. Under which circumstances would imputing with the median be more appropriate than imputing with the mean or mode?
- A
When the column has a skewed distribution
- B
When the column has a large number of categorical values
- C
When the column has a symmetric distribution
- D
When the column contains outliers
Show answer and explanation
Correct answers: A, D
Explanation
Imputing missing values with the median is particularly useful when the column has a skewed distribution or contains outliers because the median is robust to extreme values. In contrast, the mean can be heavily influenced by outliers, and the mode is more applicable to categorical data.
- A. Correct.
Imputing with the median is often more appropriate for skewed distributions because the median is less affected by extreme values compared to the mean.
- B. Incorrect.
The mode is better suited for categorical columns, so imputing with the median would not typically be relevant in this scenario.
- C. Incorrect.
For symmetric distributions, the mean and median are usually similar, so either can be used. There is no specific advantage to using the median in this case.
- D. Correct.
The median is robust to outliers, unlike the mean, making it a better choice for columns with outliers.