In the paper, we consider the problem of classical cluster algorithms applicability to the qualitative data. Assuming these data we face the problem of assigning the numerical labels to them. These labels must be coordinated with the observed joint frequencies of each combination of data categories with the best possible way. We propose a way to construct the labels. Also some estimation of the strength of non-numerical factors, based on constructed labels is proposed.

Key words: clusterization, non-numerical data, strength of statistical connection