Normalisation with respect to pattern

Abstract

The article presents a new normalisation method of diagnostic variables - normalisation with respect to the pattern. The normalisation preserves some important descriptive characteristics of variables: skewness, kurtosis and the Pearson correlation coeffcients. It is particularly useful in dynamical analysis, when we work with the whole population of objects not a sample, for example in regional studies. After proposed transformation variables are comparable not only between themselves but also across time. Then we can use them, for example, to construct composite variables.