Multivariate analysis uses statistical techniques to analyze more than 2 variables at once. It is a collection of methods used when several measurements are made on each sample. Examples are light absorbed by a sample at different wavelengths, pixels of an image, temperature and chemical composition of a sample, or temperature, pressure, and flow in a chemical plant at a certain time. Typically, one looks for (a few) latent variables or states, which are combinations of the often several hundred measurements. NORCE has long experience from such analysis, both used for classification and for prediction purposes.