Gieber, H. (2010). Multivariate statistical analysis using the R package chemometrics [Diploma Thesis, Technische Universität Wien]. reposiTUm. https://resolver.obvsg.at/urn:nbn:at:at-ubtuw:1-39732
E105 - Institut für Statistik und Wahrscheinlichkeitstherorie
-
Date (published):
2010
-
Number of Pages:
72
-
Keywords:
Multivariate Datenanalyse; Chemometrie; Statistiksoftware R
de
multivariate data analysis; chemometrics; statistics software R
en
Abstract:
In multivariate data analysis we observe not only a single variable or the relation between two variables but we consider several characteristics simultaneously. For a statistical analysis of chemical data (also called chemometrics) we have to take into account the special structure of this type of data. Classic model assumptions might not be fulfilled by chemical data, for instance there will be a large number of variables and only few observations, or correlations between the variables occur. To avoid problems arising from this fact, for chemometrics classical methods have to be adapted and new ones developed.<br />The statistical environment R is a powerful tool for data analysis and graphical representation. It is an open source software with the possibility for many individuals to assist in improving the code and adding functions. One of those contributed function packages - chemometrics implemented by Kurt Varmuza and Peter Filzmoser - is designed especially for the multivariate analysis of chemical data and contains functions mostly for regression, classification and model evaluation.<br />The work at hand is a vignette for this package and can be understood as a manual for its functionalities. The aim of this vignette is to explain the relevant methods and to demonstrate and compare them based on practical examples.