Analyzing Sequential Patterns in Retail Databases

Finding correlated sequential patterns in large sequencedatabases is one of the essential tasks in data mining since a hugenumber of sequential patterns are usually mined, but it is hard to findsequential patterns with the correlation. According to the requirementof real applications, the needed data analysis should be different. Inprevious mining approaches, after mining the sequential patterns,sequential patterns with the weak affinity are found even with a highminimum support. In this paper, a new framework is suggested for miningweighted support affinity patterns in which an objective measure,sequential ws-confidence is developed to detect correlated sequentialpatterns with weighted support affinity patterns. To efficiently prunethe weak affinity patterns, it is proved that ws-confidence measuresatisfies the anti-monotone and cross weighted support properties whichcan be applied to eliminate sequential patterns with dissimilarweighted support levels. Based on the framework, a weightedsupport affinity pattern mining algorithm (WSMiner) is suggested. Theperformance study shows that WSMiner is efficient and scalable formining weighted support affinity patterns.