Multivariate analysis – getting more out of your data

Experimental sciences often generate large amounts of complex data consisting of very many variables. Finding trends and patterns in these data sets may be difficult using standard univariate statistics.  Multivariate methods can consider all variables at once, and can be very useful to spot "hidden" patterns and relationships between variables. This presentation describes some multivariate methods for data exploration. The methods are illustrated using medical images and EPR spectra acquired by the biophysics group at UiO.