mmarizes the region,fllnction,usage,input&output, out tlle problem invjsualization,custom development,related projects;pointsface of Wbka;and presents a methOd t0 enhance the fllnctiOn Of datapreprocessiIlg of weka. W呔a usuany impresses one numerous and jumbled,that is it,there aremultifarious fI】nctjons and algorithms which is terribly complex,thinythousand lines and ejght handred files of jaVa ext.It is a significatiVe job todispan the core of Weka platfom.1mis papef will reduce the system inclassif!ier’eg.Only thirty one矗les suppOrt如e NavieBayesSimple classifieL weka belongs to untight couple£IM tools,usually spents much time Onsearching, extracting, cleanin岛 traIlsforming in tlle phase of datapreprocessing,what’s more,uⅡtight couple DM tool recIlrs to Other mettlodsto extract data,and is hard tO iIltegrate widl info咖atiOn pmcess systems.Asfar as time consuming phase of data皿Ⅱin岛data pr印mcessing takes up60%_80%Ihe whOle time,all that make it is important to improve theintelligence of data pmcessing.At last,after cOmbining with weka platf0咖and JDBCWrapper the paper analyses and designs the couple between wekaand d嚣掂1)矗se鑫魏d i珏telll{≯nl data pfep}0娌ss{鞋g. The paper benefits mose wh0 use weka platfofm Or thOse who wOrk atdevelOping pfogfanlming Of data l癌n{ng syslem, Key Wbrds:weka,Da组Min妇吕W酞a A糟hi量ec£ure,Weka CO糟,DataPreprocessing 独创性声骥 本人声明,所墨交豹学位论文是我个人在导师指导下涟行的研究工作及取褥的磷究域果。
尽本人所糕,除了文中特别加以标注稻致谢的地方外,沦文中不包含其他人已经发表或撰写过的研究成果,也不包含为获得北京交通大学或其他教学机构鲍学位或诞书两使月过的材料。
与我一起工作静同志对本{湃究所傲豹任悸贾献已奁论文中作了明确豹说明并表示了谢意。
本人签名:!童垫 日期:丛年土月卫目 关于论文使用授权的说明 本人完全了解北京交通大学有关保留、使用学位论文的规定,即:学校有权保留送交论文的复印件,允许论文被查阅和借阅;学校可以公布论文的全部或部分内容,可以采用影印、缩印或其他复制手段保存论文。
论文中所有创新和成果归北京交通大学计算机与信息技术学院所有。
未经许可,任何单位和个人不得拷贝。
版权所有,违者必究。
本人签名:!i盔 曰期:三生£年上月止日 弟1章绪论 第1