Answer: Create a feature attribution drift monitoring job. Set the sampling rate to 0.1 and the monitoring frequency to weekly.

The question requires monitoring for changes in feature interactions and understanding feature importance while minimizing costs. Feature attribution drift monitoring specifically tracks how feature importance and interactions change over time, directly addressing the concern about changing demographics affecting predictions. Option D is optimal because it uses feature attribution drift monitoring with a 0.1 sampling rate, which reduces costs by analyzing only 10% of prediction requests while still effectively detecting drift. Options A and B use basic feature drift monitoring, which only detects distribution changes in individual features but not changes in feature interactions or importance. Option C uses feature attribution drift monitoring but with a 1.0 sampling rate, which analyzes 100% of requests and is more expensive without significant benefit for drift detection. The community discussion strongly supports D (88% consensus) with upvoted comments explaining that feature attribution drift directly addresses feature interaction changes and that lower sampling rates reduce costs.

Author: LeetQuiz Editorial Team