Data Mining on Market-Basket Data

With the popularity of mobile devices, customers are able to make transactions from anywhere at anytime. These data has been digitized and collected among various market-basket databases. Mining of databases has attracted a growing amount of attention in database communities due to its wide applicability to improving marketing strategies. In this dissertation, we first study the impact of item taxonomy on the mining of transaction clusters from the retail market-basket database. Then, we take both association and taxonomy relationships into consideration for mining item clusters from the retail market-basket database. Finally, we investigate the problem of mining mobile sequential patterns from the mobile commerce market-basket database with moving patterns and purchase patterns of customers.

Explicitly, for mining transaction clusters, we devise a novel measurement, called the category-based adherence, and utilize this measurement to perform the clustering. With this category-based adherence measurement, we develop algorithm k-todes for market-basket data with the objective to minimize the category-based adherence. The distance of an item to a given cluster is defined as the number of links between this item and its nearest tode. The category-based adherence of a transaction to a cluster is then defined as the average distance of the items in this transaction to that cluster. It is shown by our experimental results, with the taxonomy information, algorithm k-todes devised in this dissertation significantly outperforms the prior works in both the execution efficiency and the clustering quality.

For mining item clusters, we devise association-taxonomy similarity and utilize this measurement to perform the clustering. With this association-taxonomy similarity measurement, we develop algorithm AT for efficiently mining item clusters. Two validation indexes based on association and taxonomy properties are also devised to assess the quality of clustering for item data. It is shown by our experimental results that algorithm AT devised in this dissertation significantly outperforms the prior works in the clustering quality as measured by the validation indexes, indicating the usefulness of association-taxonomy similarity in item data clustering.

For mining mobile sequential patterns, we devise three algorithms (algorithm TJLS, algorithm TJPT, and algorithm TJPF). Algorithm TJLS is devised in light of the concept of association rules. Algorithm TJPT is devised by taking both the concepts of association rules and path traversal patterns into consideration and gains performance improvement by path trimming. Algorithm TJPF is devised by utilizing the pattern family technique which is developed to exploit the relationship between moving and purchase behaviors. A simulation model for the mobile commerce environment is developed and a synthetic workload is generated for performance studies. It is shown by our experimental results that algorithm TJPF significantly outperforms others in both the execution efficiency and the memory saving, indicating the usefulness of the pattern family technique devised in this dissertation.