| 摘要 |
<P>PROBLEM TO BE SOLVED: To extract possibly-unknown concurrent sales information from a large amount of correlation rules in POS data analysis. <P>SOLUTION: The POS data analysis system according to the invention extracts unexpected concurrent sales information from a large amount of correlation rules by paying attention to a double hierarchical data structure like "commodity genre-commodities". In particular, a magnitude of variation of a concurrent sales tendency of commodities in a commodity genre is calculated, and a commodity genre which has a large magnitude of variation is recommended as an analysis object. Further, in a commodity genre, commodities which have a high possibility of concurrent sales is defined so as to have smaller distance between the commodities, that is, the higher the possibility the smaller the distance, and thereby the commodities are clustered. Candidate of commodities to be focused are recommended by presenting a hierarchical structure of the clustering to a user. Further, a state of concurrent sales of characteristic commodities is extracted by comparing a state of concurrent sales of commodity level with a state of concurrent sales of commodity genre level. <P>COPYRIGHT: (C)2012,JPO&INPIT |
