MVA Framework Offline Guide

Guide to the multivariate analysis (MVA) framework, a modular CMSSW-integrated framework for easy access to MVA algorithms.


  • Developer: Christophe Saout <>


The MVAComputer is a framework to allow generic discriminator computations employing multi-variate analysis techniques. The goal is an easy-to-use API and seamless CMSSW integration. The latter part is realised by storing the training data (i.e. the description of the MVA computer layout, likelihood PDF distributions, neural network configuration, ...) needed for the computation of a specific discriminator inside the CMSSW conditions database. The framework also provides an accompanying MVA trainer which can be used to create those calibrations, either from directly using the API or by using provided framework ESProducerLooper templates to facilitate running the training on EDM data files directly.


A page with an introduction and simple examples using the FWLite interface can be found here: SWGuideMVAFrameworkTutorial

To do

  • a tool to create standard XML training files for the common use cases
  • visualization tools
  • some nice examples


The SWGuideMVAComputer and SWGuideMVATrainer Wiki pages describe using the evaluator and trainer directly from within CMSSW, with interfaces to the EventSetup/CondDB and related plugin templates, etc....




Area in free format for items that the groups wish to add: meetings, to do lists, links etc

ChristopheSaout - 06 Dec 2007 created initial wiki page

Responsible: ChristopheSaout

Edit | Attach | Watch | Print version | History: r6 | r4 < r3 < r2 < r1 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r2 - 2007-12-10 - unknown

    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    CMSPublic All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback