异常数据和缺失数据的处理.pdfVIP

20 11 V o . 20 N o. 11 20 10 11 CO M PU T ER T ECHN OL O GY AN D DEV EL OPM ENT Nov. 20 10 董 艳 ( 合肥工业大学 管理学院, 安徽 合肥 230009) : , , , , , , , , , , : ; ; ; ; : T P274 : A : 1673- 629X( 2010) 11- 0225- 04 Application of Data Pre- processing Method in Mobile Telecommunication Industry DON G Y an ( Schoo of M anagement , Hef ei University of T echno ogy, Hef ei 230009 , China) Abstract : T h e main purpose of t he artic e is so ving t he dat a qua ity , in order to pre- process data f or data mining mode cust om er churn ana y sis and predict ion of a certain mobi e t e ecommunicat ion in dust ry as a subject . T he m et hod used in the paper are: dimension reduc- t ion, integration and st ructura propert ies, mu tip e imputation, discret izat ion , norma izat ion, dat a samp ing and ot her met hods t o get a comp et e, simi ar to t he rea dat a sets. In t his paper, t he processing of data cont aining a arge number of missing va ues t o t he charact eris- t ics of t he imput at ion met hods f or processing . It is inc uding t he met hod of im put at ion method of choice and using mu tip e imput at ion methods f or missing dat a correct ion. T he dat a aft er pre- processing app ied to the data mining mode improve t he eff iciency and reduce t he comp ex ity of data mining . Key words:data pre- processing ; data mining ; dat a c eaning ; mu t ip e imputat ion; missing va ue 0 ( 2) , , , ( 3) ,