Researches And Extensions Of The Basket Analysis Methods

Firstly, we introduced association rules mining technologies developed for basket analysis, the statistic loglinear model used to analyze independent and dependent qualities between factors and weighted association rules mining. Then a basket database was analyzed with SPSS's loglinear model, and the output was compared to the mined association rules. It was found that the loglinear model method precedes the association rules mining method in reflecting the association relationships between items roundly and exactly. Thereafter an algorithm of Analysis based on LogLinear model( ALL) was proposed, which gives an explorative analysis to the basket data in the first instance, then calls SPSS' s loglinear model procedure and analyzes the SPSS' s output, in the end provides easily understood analysis to users. In addition, we redefined the concepts of transaction' s weight, the weighted support and so on, and proposed an efficient weighted significant patterns tree (WSPT) algorithm based on the FPT algorithm. Finally, we utilized SPSS, some excellent algorithms and ones proposed in the paper to set up a basket analysis prototype.