RootProf: software for multivariate analysis of unidimensional profiles

Abstract

RootProfis a multi-purpose program which implements multivariate analysis of unidimensional profiles. Series of measurements, performed on related samples or on the same sample by varying some external stimulus, are analysed to find trends in data, classify them and extract quantitative information. Qualitative analysis is performed by using principal component analysis or correlation analysis. In both cases the data set is projected in a latent variable space, where a clustering algorithm classifies data points. Group separation is quantified by statistical tools. Quantitative phase analysis of a series of profiles is implemented by whole-profile fitting or by an unfolding procedure, and relies on a variety of pre-processing methods. Supervised quantitative analysis can be applied, provideda prioriinformation on some samples is provided.RootProfcan be applied to measurements from different techniques, which can be combined by means of a covariance analysis. A specific analysis for powder diffraction data allows estimation of the average size of crystal domains.RootProfborrows its graphics and data analysis capabilities from the Root framework, developed for high-energy physics experiments.