v0.12.0

✨ Added

• Drift monitoring on stratified data streams: AsymptoticStratifiedPPRM (human labels combined with proxy predictions) and AsymptoticStratifiedClassicalRM (labels only), the synthetic dataset generator generate_batched_stratified_binary_dataset (a production stream as successive batches, each independently partitioned into strata), and a new tutorial.
• Drift monitoring on clustered data streams: AsymptoticClusteredPPRM (human labels combined with proxy predictions) and AsymptoticClusteredClassicalRM (labels only), the synthetic dataset generator generate_batched_clustered_binary_dataset (a production stream as successive batches, each independently partitioned into clusters), and a new tutorial.

🔄 Changed

• Renamed the pi parameter of ASIMeanEstimator.estimate and IPWPTDMeanEstimator.estimate to sampling_probability, consistent with IPWClassicalMeanEstimator.estimate. Callers passing pi= by keyword must update; positional calls are unaffected.
• StratifiedClassicalMeanEstimator now weights strata by their share of all rows by default, labeled or not (previously, only labeled rows were counted).
• UniformSampler.sample now takes y_proxy instead of n_total, consistent with the other samplers.

💛 Contributors

Thank you to everyone who contributed to this release: @gmartinon-ed, @imerad, @pravallikamavilla