Selekcja zmiennych w analizie skupień marketingowych zbiorów danych binarnych
Selection of variables in marketing binary data cluster analysis

Summary/Abstract: In 2011 Desai proposed an interesting measure of similarity of two different values/ variants of the same variable. This measure can be easily used to assess the discrimination power of binary or multi-level nominal variable in cluster analysis. The idea is based on the fact that the smaller the similarity between e.g. 1 and 0 (treated as the binary variable values) the bigger the discrimination power of the variable. This idea was used to construct a new variable selection method for binary variables in the context of cluster analysis and for quite a broad range of binary data sets such as marketing data sets. The main advantage of the new proposal is its independence of the necessity of data grouping which is always connected with applying some grouping method and, in turn, some established number of clusters. The experiment carried out on 162 data sets shows high efficiency of the new proposal.

• Details
• Contents