Method for analyzing asymmetric clusters in spectral analysis
First Claim
1. A method for analyzing a sample comprising the steps of:
- assembling a first plurality of known samples to provide a training set of known samples;
sequentially placing each of the training set of known samples in a spectroscopic apparatus for measuring a property of the sample as a function of an observable parameter;
measuring the property of the known samples to obtain a training set of spectra at a plurality of values of the observable parameters;
storing said training set of spectra;
forming a bootstrap distribution from the training set of spectra;
measuring the property of test samples at the plurality of values of the observable parameters to obtain a set of test spectra;
storing said test spectra;
forming a univariate distribution from the bootstrap distribution, the univariate distribution comprising points of the bootstrap distribution located within a predetermined distance of a line between a center point of the bootstrap distribution and the test spectrum of the test sample;
determining skew-adjusted confidence limits from the bootstrap distribution'"'"'s center and a median of the training set median and a median of the univariate distribution; and
determining whether the test sample is a member of the sample population based on the skew-adjusted confidence limits.
1 Assignment
0 Petitions
Accused Products
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.
88 Citations
11 Claims
-
1. A method for analyzing a sample comprising the steps of:
-
assembling a first plurality of known samples to provide a training set of known samples; sequentially placing each of the training set of known samples in a spectroscopic apparatus for measuring a property of the sample as a function of an observable parameter; measuring the property of the known samples to obtain a training set of spectra at a plurality of values of the observable parameters; storing said training set of spectra; forming a bootstrap distribution from the training set of spectra; measuring the property of test samples at the plurality of values of the observable parameters to obtain a set of test spectra; storing said test spectra; forming a univariate distribution from the bootstrap distribution, the univariate distribution comprising points of the bootstrap distribution located within a predetermined distance of a line between a center point of the bootstrap distribution and the test spectrum of the test sample; determining skew-adjusted confidence limits from the bootstrap distribution'"'"'s center and a median of the training set median and a median of the univariate distribution; and determining whether the test sample is a member of the sample population based on the skew-adjusted confidence limits. - View Dependent Claims (2, 3, 4, 5, 6)
-
-
7. A method for testing a sample to determine if it is adulterated comprising the steps of:
-
assembling a first plurality of known unadulterated samples to provide a training set of known unadulterated samples; sequentially placing each of the training set of known unadulterated samples in a spectroscopic apparatus for measuring a property of the sample as a function of an observable parameter; measuring the property of the known unadulterated samples to obtain a training set of spectra at a plurality of value of the observable parameters; storing said training set of spectra of known unadulterated samples; forming a bootstrap distribution from the training set of spectra; measuring the property of the test sample at the plurality of value of the observable parameters to obtain a set of test spectra; storing said test spectra; forming a univariate distribution from the bootstrap distribution, the univariate distribution comprising points of the bootstrap distribution located within a predetermined distance of a line between the center of the bootstrap distribution and the test spectra; determining skew-adjusted confidence limits by calculating a projected difference between the bootstrap distribution'"'"'s center and a median of the training set projected onto the line, and adjusting the projected difference by a predetermined factor and a median of the univariate distribution; and determining whether the sample being tested is adulterated based on the skew-adjusted confidence limits. - View Dependent Claims (8, 9, 10, 11)
-
Specification