Prediction by regression

Despite this, predictions by regression equations that ineluded temperature were minimally better than equations that did not inelude temperature (Vecchia and Bunge, 2002b). This was probably beeause the majority of measurements were made at 30 to 35°C and contributions from other sourees of variability were not insignificant. For the databases provided here of permeability eoeffieients in animal skins, the number of measurements are too few and the temperamre variation too limited for reliable evaluation of the effeet of temperamre. 

The ultimate goal of most multivariate analyses is to develop a model to predict a property of interest. That property may be categorical or continuous. Continuous properties are modeled and predicted by regression methods—Principal Component Regression (PCR) and Partial Least Squares (PLS), in the case of Pirouette. After performing exploratory testing, the next step is to decide... 

Besides predicting categories for samples, MS e-nose instruments can also be used for determining a continuous property of samples. Continuous properties are modeled and predicted by regression methods. Details describing how SPME-MS-MVA has been used for predicting the shelf life of milk are described below (16). 

---