Basket Classifier: Fast and Optimal Restructuring of the Classifier for Differing Train and Target Samples

The common approach for constructing a classifier particle selection assumes reasonable consistency between train data samples and the target sample used particular analysis. However, may have very different properties, like energy spectra signal background contributions. We propose new method based on an ensemble of pre-trained classifiers, each trained exclusive subset, basket, total dataset. Appropriate separate adjustment separation thresholds every basket allows to dynamically adjust combined make optimal prediction with differing properties without re-training classifier. is illustrated toy example. A quality dependency number baskets also presented.