Application of the pattern-recognition method for modelling expert estimation of chromatogram quality

• Published in: Analytica chimica acta Vol. 369; no. 1; pp. 47 - 56
• Main Authors: Pirogov, A.V., Platonov, M.M., Pletnev, I.V., Obrezkov, O.N., Shpigun, O.A.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 10-08-1998
• Subjects: Chromatogram quality estimation; KNN method; Optimization criteria; KNN method; Chromatogram quality estimation; Optimization criteria
• ISSN: 0003-2670; 1873-4324
• DOI: 10.1016/S0003-2670(98)00210-4

A method of nonparametric regression, namely the method of
k-nearest neighbours, was applied in chromatography for creating an estimation criterion of chromatogram quality. A training set was constructed using expert estimation of a number of model and real chromatograms, and the method was learned to estimate chromatograms as a single image. The parameters for chromatogram estimation were selected, and their variance weights were determined. The proposed approach is compared with the existing criteria, and its advantages are shown. The method was successfully applied for searching for the optimum conditions for the ion-chromatographic determination of inorganic anions.