Approach to visualisation of evolving association rule models

Visualization of evolving data mining models can allow good insight for a data analyst about these models and their changes in time. This paper presents our approach to visualization of evolving association rule models. This approach is based on graph visualization where nodes of the graph represent itemsets, and edges represent association rules. We show how evolving models, produced by data mining algorithms, are stored in the knowledge base, and then how they can be filtered and visualized. Two ways of graph based visualization of evolving models are shown using force based layout algorithms - local and global layout. Experiments with both are illustrated with emphasis on the latter.