Method of cluster making for a multivariate analysis.

MITSUI Toshiyuki, OKUYAMA Syuji, HIDA Minemasa

doi:10.2116/bunsekikagaku.52.239

A qualitative analysis involving multivariate analysis does not have a clear standard on which to base the results. Any difference in the qualitative analytical results by individual analysts raises doubt concerning the reliability of the qualitative analysis in which a multivariate analysis is used. The qualitative analysis for identification is evaluated as follows. First, the obtained 5∼15 data from measuring one sample is added to unknown cluster belonging to a sample (unknown sample) to identify the difference, and calculated by a principal component analysis for removing the outlier in unknown samples. Next, the cluster analysis is examined. When the Euclid distance of the unknown sample cluster is smaller than that of the data cluster, the unknown sample cluster is judged as one cluster. Next, the relations between the unknown samples are calculated by soft independent modeling of the class analogies method (SIMCA method) based on the result of the cluster analysis. The unknown samples are calculated by a principal component analysis based on the result of the SIMCA method. Finally, the cluster belonging of the individual unknown sample is specified from the result of the cluster analysis and the principal component analysis. The unknown samples are accurately made into a cluster by such a method.<br>

Method of cluster making for a multivariate analysis.

Bibliographic Information

Search this article

Description

Journal

Citations (2)*help

References(12)*help

Keywords

Details 詳細情報について

Export

Report a problem