TMVA- the toolkit for multivariate data analysis
J. Stelzer · 2009
Multivariate classification methods based on machine learning techniques play a fundamental role in today's high-energy physics analyses.In a time of ever larger datasets with an ever smaller signal fraction it becomes increasingly important to use all the features of signal and background that are present.Not only the one dimensional variable projections of the signal must be evaluated against the background, in particular the information present in the variable correlations must be explored.The possible complexities of the data distributed in a high-dimensional space are difficult to disentangle manually and without the help of automata.TMVA is a toolkit which implements a large variety of multivariate classification algorithms.TMVA provides a framework to simultaneously train and evaluate any chosen set of classifiers, and giving means to compare the performance on any dataset for any given problem.