Multivariate Data Analysis with TMVA
In high-energy physics, with the search for ever smaller signals in ever larger data sets, it has become essential to extract a maximum of the available information from the data. Multivariate classification methods based on machine learning techniques have become a fundamental ingredient to most HEP analyses. Also the multivariate classifiers themselves have significantly evolved in recent years. Statisticians have found new ways to combine many input variables into powerful classifiers to further gain in performance over traditional multivariate techniques.
Integrated into the analysis framework ROOT, TMVA is a toolkit which hosts a large variety of multivariate classification algorithms, and - as a new feature - TMVA has been extended to multivariate regression analysis. All methods are embedded in a framework capable of handling a coherent pre-processing of the data and evaluation of the output, thus allowing a simple and convenient use of multivariate techniques, and objective performance assessment.
The seminar provides an overview of the multivariate approaches implemented in TMVA, and describes the usage and features of the TMVA package.
The experimental particle physics seminar series invites speakers from all over Europe to discuss the latest developments at the LHC, accelerator and non-accelerator based neutrino physics, hardware R&D and astroparticle physics. .