METHOD FOR ANALYZING LARGE-SIZED DISK RESIDENT DATA SET BY USING PATIENT RULE INDUCTION METHOD INSIDE COMPUTER SYSTEM, INSIDE PARALLEL ARCHITECTURE COMPUTER AND INSIDE SYMMETRICAL MULTIPROCESSOR ARCHITECTURE COMPUTER

摘要

PROBLEM TO BE SOLVED: To provide a method for analyzing a large-sized disk resident data set by using a patient rule induction method (PRIM). SOLUTION: Inside the PRIM, a relation data chart 50 containing a continuous attribute 52, a non-continuous attribute 54, a material parameter and a cost attribute 56 is received first. The cost attribute 56 indicates a cost output value on the basis of continuous and non-continuous attribute values as input, a super rectangle surrounding a multi-dimensional space defined by the continuous and non-continuous attribute values is formed and the continuous and non-continuous attribute values are displayed as points within the multi-dimensional space. Until the total number of the points within the super rectangle becomes equal to a two-dimensional parameter 58, the plural points are removed from the plural points on the basis of the average of the cost output value along the edge of the super rectangle. Until the total of the cost output values in the entire multi-dimensional space surrounded by the super rectangle changes, the removed non-continuous and continuous attribute values are added along the edge of the super rectangle.