Model Validation in Near Infrared Analysis: A Comparison of Different Validation Methods

Near-infrared spectroscopy (NIRS) is widely used in different fields by virtue of high speed, low cost and its ability to record spectra for solid and liquid samples without any pre-treatment. Associated with chemometrics, it becomes a powerful tool for in-line or off-line quantitative analysis. It seems easy to build NIR calibration models, however, some of them look good at first sight but prove unreliable when used in practical. Thus, how to do properly model validation becomes very important. In this study, different cross validation methods are compared by using some real NIR data sets and the results are discussed.