| abstract |
The multiple linear regression approach typically used in near-infrared spectrometry yields equations in which any amount of reflectance at the analytical wavelengths leads to a corresponding composition value. As a result, when the sample contains a component not present in the training set, erroneous composition values can arise without any indication of error. There is described a method of detecting "false" samples by constructing a multi-dimensional form in space using reflectance values of samples in a training set at a number of wavelengths. A new sample is projected into this space and a confidence test is executed to determine whether the new sample is part of the population from which the training set was drawn. The method relies on few assumptions about the structure of the data; therefore, deviations from assumptions do not affect the results of the confidence test. |