Study on Attribute Reduction Method in Sample
Information Processing
Xia Kewen,Shen Junyi,Li Changbiao
(School of Electronics and Information Engineering,Xi'an Jiaotong University,Xi'an
710049,China)
Abstract:In order to eliminate redundancies and incompatibilities in samples,extract
the key information, and so on,the condition attribute significance was defined
scientifically based on two approximation qualities of rough set according to the
characteristics of sample information and the thought of information
granularity.Moreover,a simple effective method of attribute reduction was presented,the
main process was to determine the information core,calculate significance of dispensable
condition attribute and obtain the relative attribute reduction set.For the continuous
attribute processing, a speedy discretization algorithm was presented based on fuzzy
similar-ratio principle which can eliminate the fuzzy noise.The typical example
calculation and the actual application in oil-water formation recognition system show that
discernment accuracy of the attribute reduction method can approach more than 90%,and the
application effect is very notable.
Keywords:attribute reduction;sample information;approximation quality;continuous attribute
discretization; fuzzy similar-ratio