Inventors:
Taiki Sakata - Sunnyvale CA, US
Charles Herman - Dublin CA, US
Steven Dolins - Highland Park IL, US
International Classification:
G06F007/00
Abstract:
A method of analyzing data contained in a very large database or data warehouse includes applying first a patient rule induction method with selected input variables and an output variable to develop a region containing a subset of database records having a highest average output value. Then, the subset of records may be aggregated, sorted, compared, or new measures computed, preferably using an on line analytical processing function, for an pattern analysis of the subset records. An alternative is to apply a weighted item set function to the subset to produce a reduced subset, and then applying the reduced subset to on line analytical processing for pattern analysis.