CERTIFIED-MACHINE-LEARNING-PROFESSIONAL · Question #54
CERTIFIED-MACHINE-LEARNING-PROFESSIONAL Question #54: Real Exam Question with Answer & Explanation
The correct answer is C: Mode, number of unique values, and percentage of missing values. Option C is correct because monitoring categorical feature drift requires tracking multiple complementary statistics: mode detects shifts in the most frequent category, number of unique values reveals if new categories appear or old ones disappear, and percentage of missing value
Question
Which of the following is a simple statistic to monitor for categorical feature drift?
Options
- AMode
- BNone of these
- CMode, number of unique values, and percentage of missing values
- DPercentage of missing values
- ENumber of unique values
Explanation
Option C is correct because monitoring categorical feature drift requires tracking multiple complementary statistics: mode detects shifts in the most frequent category, number of unique values reveals if new categories appear or old ones disappear, and percentage of missing values catches data pipeline issues or schema changes - together they give a complete picture of how a categorical distribution is changing over time.
The single-statistic options (A, D, E) are each incomplete on their own - mode alone misses cardinality changes, missing value rate alone misses distributional shifts, and unique value count alone misses which categories dominate. Option B is wrong because useful statistics for categorical drift absolutely exist.
Memory tip: Think of categorical drift monitoring as a "Who, How Many, How Complete" check - who is dominant (mode), how many categories exist (unique values), and how complete is the data (missing values). All three together, never just one.
Community Discussion
No community discussion yet for this question.