An expert system for the evaluation and interpretation of interlaboratory comparisons

Abstract Danzer, K., Wank, U. and Wienke, D., 1991. An expert system for the evaluation and interpretation of interlaboratory comparisons. Chemometrics and Intelligent Laboratory Systems , 12: 69–79. An expert system is presented which performs complex analyses of interlaboratory tests of several types. The system is arranged in three parts, in relation to the sequence of a statistical analysis (computation of the series means, computation of the total mean, evaluation of the interlaboratory test). These parts are connected but can also be used independently of each other, for instance for a common statistical evaluation of a measuring series. The expert system is a collection of essential parametric and robust statistical procedures. The selection of the most suitable test is governed by the properties of the actual data. The evaluation algorithm is guided by decision rules working in the background. In this way test results can be explained or interpreted and subsequent steps of an evaluation will be proposed to the user. An experienced user can also select other statistical tests from the appropriate menu, or can ignore the advice completely.