Multivariate Calibration

An interesting variable may depend on several descriptors simultaneously. In order to make a quantitative prediction of this variable, one has to set up a multivariate model, which establishes a quantitative relationship between the descriptor variables x_i and the target (response) variable y.

There are several methods available for building a multivariate model. Here's a list of the more important ones:

• multiple linear regression (MLR)
• principal component regression (PCR)
• partial least squares regression (PLS)
• artificial neural networks (ANN)