rootpy.stats.histfactory.split_norm_shape

rootpy.stats.histfactory.split_norm_shape(histosys, nominal_hist)[source]

Split a HistoSys into normalization (OverallSys) and shape (HistoSys) components.

It is recommended to use OverallSys as much as possible, which tries to enforce continuity up to the second derivative during interpolation/extrapolation. So, if there is indeed a shape variation, then factorize it into shape and normalization components.