"Low cost" method is the key here, computationally simple on a spark/delta back end, as summary stats are compiled/updated automatically...won't require a re-index/shuffle. Answer is B.

Monitoring summary statistics involves tracking basic statistical measures such as mean, median, variance, and standard deviation over time. By comparing these statistics between different time periods or datasets, you can detect significant changes that may indicate feature drift. Summary statistics trends also meets the simple and low-cost method requirements.

E is answer

Monitoring summary statistics trends over time is a simple and low-cost method of monitoring numeric feature drift. It involves tracking basic statistical metrics such as mean, median, standard deviation, etc., and observing how they change over time. This method provides insights into whether the distribution of the feature values is shifting, which could indicate drift.

B. Summary statistics trends Monitoring changes in summary statistics such as mean, median, standard deviation, and other relevant metrics over time can provide valuable insights into numeric feature drift. This method is simple, easy to implement, and does not require sophisticated statistical tests.

https://www.evidentlyai.com/blog/data-drift-detection-large-datasets Is it this?