Home

Introduction

These pages document the RooStats / RooFit - based software tools used for statistical analysis within the Higgs PAG - combine.

Combine provides a command line interface to many different statistical techniques available inside RooFit/RooStats used widely inside CMS.

The package exists in GIT under https://github.com/cms-analysis/HiggsAnalysis-CombinedLimit

For more information about GIT and its usage in CMS, see http://cms-sw.github.io/cmssw/faq.html

The code can be checked out from GIT and compiled on top of a CMSSW release that includes a recent RooFit/RooStats

Table of Contents

• Getting started
  • For Users
  • For Developers
  • Combine Tool
• Setting up the analysis
  • Preparing the datacard
    • Counting experiment
    • Shape Analysis
      • Binned
      • Unbinned (or parametric)
    • Beyond simple datacards
      • Rate Parameters
      • Manipulation of Nuisance Parameters
      • Combining Cards
      • Automatic production of datacards and workspaces
  • Physics Models
    • Model Building
      • example: Multi-signal model
      • example: Two-hypothesis model
      • example: Interference model
  • Automatic statistical uncertainties
• Running Combine
  • How to run the Tool
    • Common Command Line Options
      • Generic Minimizer Options
      • Output from Combine
    • Toy data Generation
      • Asimov datasets
      • Nuisance parameter Generation
      • Loading Snapshots
    • combineTool for Job Submission
      • Submission to LSF Batch
      • Submission to the Grid
  • Common statistical methods
    • Asymptotic Frequentist Limits
      • Blind Limits
      • Splitting Points
    • Asymptotic Significances
      • Compute the Observed Significance
      • Computing the Expected Significance
    • Bayesian Limits and Credible regions
      • Observed (simple models)
      • Observed (arbitrary models)
      • Expected Limits
      • Multidimensional Credible Regions
    • Computing Limits with toys
      • Simple Models
      • Complex Models
    • Computing Significances with toys
      • Observed Significance
      • Expected Significance
    • Goodness of Fit
    • Channel Compatibility
    • Likelihood Fits and Scans
      • Useful options for Scans
      • Using best fit Snapshots
    • Feldman Cousins
  • Advanced use cases
    • Fitting Diagnostics
      • Fit parameter uncertainties
      • Pre/Post nuisance pars and pulls
      • Normalizations
      • Plotting
      • Toy-by-toy Diagnostics
    • Scaling constraints
    • Nuisance Parameter Impacts
    • Channel Masking
    • RooMultiPdf - Bias studies
    • RooSplineND - Multidimensional splines
    • RooParametricHist - GammaN for shapes
• Useful Links
• FAQ