Databricks Machine Learning Associate Question 177
Single answerYou are working with a dataset for a classification problem, and one of the categorical features has missing values. When would it be appropriate to replace the missing values in this feature with the mode value?
- A
When the feature has a low cardinality and the mode represents the most frequent category.
- B
When the feature has a high cardinality and the missing values are uniformly distributed.
- C
When the feature is numerical and follows a normal distribution.
- D
When the missing values in the feature are rare and distributed randomly.
Show answer and explanation
Correct answer: A
Explanation
Replacing missing values with the mode is most appropriate for categorical features with low cardinality, where the mode effectively represents the most frequent category. This approach minimizes distortion and retains the feature's interpretability. It is less suitable for high cardinality features or numerical data, as it may introduce bias or distort the statistical properties.
- A. Correct.
Replacing missing values with the mode is appropriate when the feature has low cardinality (few unique categories) and the mode accurately represents the most frequent category. This ensures the imputation retains meaningful information.
- B. Incorrect.
Replacing missing values with the mode in high cardinality features may introduce bias, as the mode might not represent the data distribution well in such cases.
- C. Incorrect.
The mode is not a suitable replacement for missing values in numerical features, especially when they follow a normal distribution, as it may distort the statistical properties of the feature.
- D. Incorrect.
Randomly distributed missing values do not inherently justify replacing them with the mode. The appropriateness of using the mode depends on the feature's characteristics, not just the distribution of missing values.