When performing an analysis there are important procedures that usually have to be performed for an analysis. Hence tensorwaves could standardize these to some degree. The issue serves more as a discussion platfrom for now. @wgradl @redeboer @Leongrim @sebastianJaeger please feel free to extend and modify this issue. The procedures are:
- input output checks
- statistical validation. This means performing N independent full toy MC data analysis. One full toy MC data analysis consists of generating a toy MC data sample, and fitting the sample. All in all this returns N fit results, which can be combined into residual distributions for the fit parameters.
- Bootstrapping could be used here, to minimize the number of data samples that need to be created