Abstract
In multivariate analysis, the means, the variances and the correlation coefficients are widely used. As there are many methods of cluster analysis, it is difficult to chose a proper method. The probability ellipses which have the mean, the variance and the correlation coefficient are drawn in a bivariate control chart. The clustering are made by use of the relative situation and the size of the probability ellipses on that chart. Therefore, this cluster analysis contains the basic statistics which are the means, the variances and the correlation coefficients.
In this cluster analysis, the nearest neighbour method is used as the distance between the clusters. And the distance between two samples is defined by two methods using probability ellipse. The one is based on the thought that two ellipses adjoin and the other is, that one ellipse adjoin the center of other ellipse. For the latter, as the distance is asymmetric, we can find difference between characters of two samples' neighbourhoods.
Moreover, as an application of this method, cluster analysis of EEG is studied. This method is applied to 2-dimentional display of the frequency vs. the amplitude. This cluster analysis have basic statistics and visual representation.
