Incremental algorithms based on metric for finding reduct in dynamic decision tables

Abstract

Feature  selection  is  a  crucial  problem need to be effectively solved in knowledge discovery in  databases  because  of  two  basic  reasons:  to minimize  cost  and  to  accurately  classify  data. Feature selection using rough set theory also called attribute  reduction  have  attracted  much  attention from  researchers  and  many  results  are  gained. However,  attribute  reduction  in  dynamic  databases is still in the first stage. This paper focus on develop incremental  methods  and  algorithms  to  derive reducts  hiring  a  distance  measure  when   adding, deleting or updating objects. Since not re-implement the  algorithms  on  the  varied   universal  set,  our algorithms  significantly  reduce  the  complexity  of implementation time.