{"id":"arxiv/2609.26652","url":"https://arxiv.org/abs/2609.26652","tags":["math.ST","stat.ML"],"score":0,"title":"Context-Adaptive Thresholding for Conditionally Representative Monitoring and Classification","domain":"math.ST","source":"arxiv","authors":["Ansgar Steland"],"reviews":0,"subject":"math.ST","summary":"Commonly, classifiers and monitoring procedures are trained from labeled data by optimizing an objective such as the misclassification rate. This may lead to unrepresentative conditional distributions of the outcome (the labels) given important external variables, different from the conditional laws in the population. We show how to modify any given threshold-type classifier resp. monitoring rule to achieve representative conditional label prediction by using adapting the threshold to a covariate $Z$ (the context) to distribute sensitivity while maintaining the false alarm rate. In case that…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-22","reactions":{}}