Showing papers by "Klaus Danzer published in 1991"
TL;DR: In this article, a multivariate auto-and cross-correlation method for handling correlation problems is described, which allows the calculation of relationships between two functions including one or several variables with respect to changes in time or distance.
22 citations
TL;DR: In this article, the authors defined the ideal information amount which is determined by the ideal spatial resolving power A* that is theoretically attainable by actual scanning or mapping, compared with it, the real information amount of distribution-analytical results are determined by a practical spatial resolution power and therefore, among others, by the number of measuring points.
Abstract: The information of distribution analyses considerably exceeds that of integral analyses because of the spatial resolution. Distribution-analytical procedures are characterized by the ideal information amount which is determined by the ideal spatial resolving power A* that is theoretically attainable by actual scanning or mapping. Compared with it, the real information amount of distribution-analytical results are determined by a practical spatial resolving power and therefore, among others, by the number of measuring points. In dependence on the stochastical properties of multi-phase materials an optimum area can be specified for distribution-analytical investigations with a maximum information content. Methods of testing homogeneity can be characterized by means of the information theory in dependence on the arrangement of measuring points, on the evaluation of concentration or measuring values, and on their univariate, multiple or multivariate character.
5 citations
TL;DR: The least median squares regression has proven to be an advantageous method for the description of widely varying data sets which are often obtained as a result of analytically investigated environmental problems as mentioned in this paper.
Abstract: Methods of robust statistics have successfully been applied to eliminate outliers, in the classical sense, in a number of cases. The least median squares regression has proved to be an advantageous method for the description of widely varying data sets which are often obtained as a result of analytically investigated environmental problems. The least median squares regression allows the computation of correlations between the plant lead contents and the soil lead contents, the traffic density or the distance from the road. These relations between the plant lead contents on the one hand and the soil lead contents or the traffic density on the other hand can help to interprete the uptake mechanisms, caused by lead immissions. The application of reweighted squares regression seems to be less useful. Because of the elimination of extreme measuring points, the statistical reliability of the regression models obtained is not as high as that of the least median squares regression. Robust estimations of the expected plant lead contents can be obtained by using the model of multiple least median squares regression.
2 citations
TL;DR: In this article, a supervised learning procedure for the detection and characterization of classes which overlap or are located within other classes is described for an environmental problem involving an industrial source of emission and its immission effects.
2 citations
01 Jan 1991
TL;DR: Three software packages for specialized statistical analysis in the chemical laboratory are presented and two of them are new and innovative approaches to statistical analysis are presented.
Abstract: Three software packages for specialized statistical analysis in the chemical laboratory are presented.
1 citations