\vspace{-1cm}

\begin{flushright}

{\sf\em arXiv:physics/0703039 [Data Analysis, Statistics and Probability]} \\

{\sf\em CERN-OPEN-2007-007} \\

{\sf\em TMVA version \TMVAVersion} \\

{\sf\em \today} \\

\def\UrlFont{\sf\em}

\url{https://root.cern/tmva}

\end{flushright}

\def\UrlFont{\rm}

\def\miniPageOffset{0.2cm}

\def\miniPageWidth{13.5cm}

\vspace*{\stretch{20}}

\HRule

\begin{flushleft}

\hspace{\miniPageOffset}\begin{minipage}{\miniPageWidth}

{\sf\Huge\bfseries\boldmath TMVA 4} \\[0.2cm]

{\sf\Large\bfseries\boldmath Toolkit for Multivariate Data Analysis with ROOT} \\[1cm]

{\sf\Huge\bfseries\boldmath Users Guide}

\end{minipage}

\end{flushleft}

\HRule

\vspace{2.0cm}

%\begin{flushleft}

\begin{flushright}

% \hspace{\miniPageOffset}\begin{minipage}{\miniPageWidth}

{\sf\Large

K.~Albertsson,

~S.~Gleyzer,

~A.~Hoecker,

~L.~Moneta,

~P.~Speckmayer,\\

~J.~Stelzer,

~J.~Therhaag,

~E.~von Toerne,

~H.~Voss,

~S.~Wunsch

}

% \end{minipage}

\vspace{1.2cm}

% \hspace{\miniPageOffset}\begin{minipage}{\miniPageWidth}

% {\sf\em\large Contributed to TMVA have:} \\[0.4cm]

% {\sf\large

% M.~Backes,

% ~A.~Bagoly,

% ~A.~Bevan,

% ~A.~Carnes,

% ~T.~Carli,

% ~O.~Cohen,

% ~A.~Christov,%\\[0.1cm]

% ~D.~Dannheim,

% ~K.~Danielowski,

% ~S.~Henrot-Versill\'e,

% ~V.~Ilievski,

% ~M.~Jachowski, %\\[0.1cm]

% ~K.~Kraszewski,

% ~A.~Krasznahorkay Jr.,

% ~M.~Kruk,

% ~Y.~Mahalalel,

% ~A.~Moudgil,

% ~R.~Ospanov, %\\[0.1cm]

% ~S.~Pfreundschuh,

% ~X.~Prudent,

% ~A.~Robert,

% ~C.~Rosemann,

% ~P.~Seyfert,

% ~D.~Schouten,

% ~R.~Selvam,

% ~S.~Shekkar,

% ~M.~Stergiadis,

% ~T.~Stevenson,

% ~F.~Tegenfeldt,

% ~A.~Voigt,

% ~K.~Voss,

% ~M.~Wolter,

% ~O.~Zapata~Mesa,

% ~A.~Zemla,

% ~J.~Zhong

% }

% \end{minipage}

\vspace*{\stretch{1}}

\end{flushright}

%\end{flushleft}

\vfill

\thispagestyle{empty}

\newpage

% authors

\def\UrlFont{\sf}

{\small\sf

{\sf\bfseries Abstract} ---

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 analyses. Also the multivariate classifiers

themselves have significantly evolved in recent years. Statisticians

have found new ways to tune and to combine classifiers to further gain

in performance. Integrated into the analysis framework ROOT, TMVA is

a toolkit which hosts a large variety of multivariate classification

algorithms. Training, testing, performance evaluation and application of

all available classifiers is carried out simultaneously via user-friendly

interfaces. With version 4, TMVA has been extended to multivariate

regression of a real-valued target vector. Regression is invoked through

the same user interfaces as classification. TMVA 4 also features

more flexible data handling allowing one to arbitrarily form

combined MVA methods. A generalised boosting method is the first

realisation benefiting from the new framework.

}

\vspace{0.5cm}

\begin{center}

{\small\sf

{\sf\bfseries TMVA \TMVAVersion\ -- Toolkit for Multivariate Data Analysis with ROOT} \\

Copyright\index{Copyright}

\copyright\ 2005-2018, Regents of

CERN (Switzerland),

DESY (Germany),

MPI-Kernphysik Heidelberg (Germany),

University of Bonn (Germany),

and University of Victoria (Canada). \\

BSD license: \url{https://root.cern/tmva/license}.

\hspace{0.5cm} \\

{\sf\bfseries Authors:} \\

Kim Albertsson (LTU/CERN, Sweden/Switzerland) \emailsf{<kim.albertsson@cern.ch>},\\

Sergei Gleyzer (CERN, Switzerland) \emailsf{<sergei.gleyzer@cern.ch>}, \\

Andreas Hoecker (CERN, Switzerland) \emailsf{<andreas.hoecker@cern.ch>}, \\

Lorenzo Moneta (CERN, Switzerland) \emailsf{<lorenzo.moneta@cern.ch>}, \\

Peter Speckmayer (CERN, Switzerland) \emailsf{<peter.speckmayer@cern.ch>}, \\

J\"org Stelzer (CERN, Switzerland) \emailsf{<joerg.stelzer@cern.ch>},\\

Jan Therhaag (Universit\"at Bonn, Germany) \emailsf{<therhaag@physik.uni-bonn.de>}, \\

Eckhard von Toerne (Universit\"at Bonn, Germany) \emailsf{<evt@physik.uni-bonn.de>},\\

Helge Voss (MPI f\"ur Kernphysik Heidelberg, Germany) \emailsf{<helge.voss@cern.ch>},\\

Stefan Wunsch (CERN, Switzerland) \emailsf{<stefan.wunsch@cern.ch>}, \\

\hspace{0.5cm} \\

{\sf\bfseries Contributions from:} \\

Moritz Backes, % (Geneva University, Switzerland) \emailsf{moritz.backes@cern.ch},\\

Attila Bagoly,

Adrian Bevan,

Tancredi Carli, % (CERN, Switzerland) \emailsf{<tancredi.carli@cern.ch>}, \\

Andrew Carnes

Or Cohen, % (CERN, Switzerland and Technion, Israel) \emailsf{<or.cohen@cern.ch>}, \\

Asen Christov, % (Universit\"at Freiburg, Germany) \emailsf{<christov@physik.uni-freiburg.de>}, \\

Krzysztof Danielowski, % (IFJ and AGH/UJ, Krakow, Poland) \emailsf{<Krzysztof.Danielowski@cern.ch>}, \\

Dominik Dannheim, % (CERN, Switzerland) \emailsf{<Dominik.Dannheim@cern.ch>}, \\

Sophie Henrot-Versill\'e, % (LAL Orsay, France) \emailsf{<versille@lal.in2p3.fr>}, \\

Vladimir Ilievski,

Matthew Jachowski, % (Stanford University, USA) \emailsf{<jachowski@stanford.edu>}, \\

Kamil Kraszewski, % (IFJ and AGH/UJ, Krakow, Poland) \emailsf{<kamil.bartlomiej.kraszewski@cern.ch>}, \\

Attila Krasznahorkay Jr., % (CERN, CH, and Manchester U., UK) \emailsf{<Attila.Krasznahorkay@cern.ch>}, \\

Maciej Kruk, % (IFJ and AGH/UJ, Krakow, Poland) \emailsf{<maciej.mateusz.kruk@cern.ch>}, \\

Yair Mahalale, %l (Tel Aviv University, Israel) \emailsf{<yair@mahalalel.com>}, \\

Abhinav Moudgil, % (IIIT Hyderabad, India) \emailsf{<abhinav.moudgil@research.iiit.ac.in>}, \\

Rustem Ospanov, % (University of Texas, USA) \emailsf{<rustem@fnal.gov>}, \\

Simon Pfreundschuh,

Xavier Prudent, % (LAPP Annecy, France) \emailsf{<prudent@lapp.in2p3.fr>}, \\

Arnaud Robert, % (LPNHE Paris, France) \emailsf{<arobert@lpnhe.in2p3.fr>}, \\

Christoph Rosemann, % (DESY, Germany) \emailsf{<christoph.rosemann@desy.de>}, \\

Doug Schouten, % (S. Fraser U., Canada) \emailsf{<dschoute@sfu.ca>}, \\

Ravi Selvam,

Paul Seyfert,

Saurav Shekkar,

Peter Speckmeyer,

Manos Stergiadis,

Tom Stevenson,

Fredrik Tegenfeldt, % (Iowa University, USA) \emailsf{<fredrik.tegenfeldt@cern.ch>}, \\

Alexander Voigt, % (CERN, Switzerland) \emailsf{<alexander.voigt@cern.ch>}, \\

Kai Voss, % (University of Victoria, Canada) \emailsf{<kai.voss@cern.ch>}, \\

Marcin Wolter, % (IFJ PAN Krakow, Poland) \emailsf{<marcin.wolter@ifj.edu.pl>}, \\

Andrzej Zemla, % (IFJ PAN Krakow, Poland) \emailsf{<zemla@corcoran.if.uj.edu.pl>}, \\

Omar Zapata Mesa

Jiahang Zhong % (Academia Sinica, Taipei, Taiwan) \emailsf{<Jiahang.Zhong@cern.ch>}, \\

\\

\hspace{0.5cm} \\

and valuable contributions from many users, please see acknowledgements.

}

\end{center}

\thispagestyle{empty}

\newpage

%%% Local Variables:

%%% mode: latex

%%% TeX-master: "TMVAUsersGuide"

%%% End: