Fachartikel
Patterns of Associations in Finite Sets of Items

Patterns of Associations in Finite Sets of Items

Mining association rules is well established in quantitative business research literature and makes up an up-and-coming topic in marketing practice. However, reducing the analysis to the assessment and interpretation of a few selected rules does not provide a complete picture of the data structure revealed by the rules.
This paper introduces a new approach of visualizing relations between items by assigning them to a rectangular grid with respect to their mutual association. The visualization task leads to a quadratic assignment problem and is tackled by means of a genetic algorithm. The methodology is demonstrated by evaluating a set of rules describing marketing practices in Russia.