What is A/B testing in classification model deployment?

A/B testing compares two classification models in production to measure performance differences.

Traffic is split between model A (baseline) and model B (new model). Metrics like CTR, precision, recall, and conversion rate are monitored. Statistical significance testing ensures observed improvements are not due to randomness.