Servicio de Información Comunitario sobre Investigación y Desarrollo - CORDIS

Analysis of large datasets by means of correlation analysis and principal component analysis

The identification of the single spectra out of a large datasets obtained e.g. from time resolved experiments can be analyzed by the combination of the 2D correlation analysis and Principal Component Analysis. As a first step the dataset is analyzed by means of Moving window 2D correlation analysis with an appropriate size of the window as described in literature to obtain the spectral position of interest. This is especially interesting on cyclic perturbations in experiments.

Afterwards to identify the spectra of the pure components as well as the concentration profile of these on the dataset the advanced least square fitting procedure is applied. The input concentration matrix is determined by mean of Sample-Sample correlation, which gives a suitable concentration profile to identify the species. Furthermore the resolving of complex spectra with respect to the unsynchronized behaviour (which species is formed prior or precede compared to each other).

Reported by

Institut für Technische Chemie
Síganos en: RSS Facebook Twitter YouTube Gestionado por la Oficina de Publicaciones de la UE Arriba