Multivariate calibration. II. Chemometric methods

摘要: Abstract In this outline of new approaches to multivariate calibration in chemistry the following topics are treated: Advantages over conventional univariate calibration: detect and eliminate selectivity problems. Multivariate methods based on selection some variables vs. data compression all variables. Direct indirect pure constituents or known samples for calibration? Calibration by physical modelling: Beer's law. Use law controlled natural generalized least-squares fit best linear predictor. Extending handle unknown factor principal component regression partial regression. Methods detecting abnormal (outliers). Pre-treatments linearize data.