/reweighting_tutorial

Tutorial on BDT reweighter

Primary LanguageJupyter Notebook

#Tutorial on BDT reweighter

We are considering reweighting problem (introduction to this problem can be found in the post) and show how parameters of classifier/regressors tuning affects on the reweighting rule reconstruction.

Tutorial notebook consists of several parts:

  • Neural network parameters: demonstration how these parameters influence reweighting rule;
  • Variance and bias errors discussion
  • BDT reweighter tuning
  • real use-case in high energy physics (HEP): reweighting problem for sPlot data, when weights (that can be negative) are defined for the target distribution.

In the experiments we use hep_ml library and carl library.