資源描述:
《乳腺癌分類器及數(shù)據(jù)樣本驗證(python)》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫。
1、乳腺癌分類器及數(shù)據(jù)樣本驗證ByTobyQQ:231469242歡迎愛好者交流,并改進代碼數(shù)據(jù)下載地址ucimachinelearing/breastcancer詞匯:Malignancy惡性biopsy活組織檢查benign良性的diagnosis診斷periodicexamination定期檢查ClumpThickness腫塊厚度UniformityofCellSize細(xì)胞大小的均勻性UniformityofCellShape細(xì)胞形狀的均勻性MarginalAdhesion邊緣粘SingleEpithelialCellSize單上
2、皮細(xì)胞的大小BareNuclei裸核BlandChromatin乏味染色體NormalNucleoli正常核Mitoses有絲分裂背景知識isconsinBreastCancerDatabase(WBCD)January8,1991RevisedNomeber3,1994ThisisadescriptionoftheWisconsinBreastCancerDatabase,collectedbyDr.WilliamH.Wolberg,UniversityofWisconsinHospitals,Madison.Theactuald
3、atabaseiscontainedinanotherfile(datacum).SampleswerecollectedperiodicallyasDr.Wolbergreportedhisclinicalcases.Thedatabasethereforereflectsthischronologicalgroupingofthedata.Thesamplesconsistofvisuallyassessednuclearfeaturesoffineneedleaspirates(FNAs)takenfrompatients'b
4、reasts.Eachsamplehasbeenassigneda9-dimensionalvector(attributes3to9below)byDr.Wolberg.Eachcomponentisintheinterval1to10,withvalue1correspondingtoanormalstateand10toamostabnormalstate.Attribute1issamplenumber,whileattribute2designateswhetherthesampleisbenignormalignant.
5、Malignancy惡性isdeterminedbytakingasampletissuefromthepatient'sbreastandperformingabiopsyonit.Abenign良性的diagnosis診斷isconfirmedeitherbybiopsy活組織檢查orbyperiodicexamination定期檢查,dependingonthepatient'schoice.Allgroupsareinthesamefile.Wehaveseparatedthegroups感謝Wisconsin醫(yī)學(xué)院的wil
6、liamH.Wolberg博士提供乳腺癌數(shù)據(jù)樣本。所欲數(shù)據(jù)來自真實臨床案例,每個案例有9個屬性這就是判斷乳腺癌的9個屬性(翻譯非全部準(zhǔn)確)FieldAttribute1Samplecodenumber(病人ID)2Class:2forbenign,4formalignant(惡性或良性分類)3ClumpThickness腫塊厚度4UniformityofCellSize細(xì)胞大小的均勻性5UniformityofCellShape細(xì)胞形狀的均勻性6MarginalAdhesion邊緣粘7SingleEpithelialCellSize
7、單上皮細(xì)胞的大小8BareNuclei裸核9BlandChromatin乏味染色體10NormalNucleoli正常核11Mitoses有絲分裂數(shù)據(jù)樣本示例說明:病人ID惡性或良性(2是良性,4是惡性)剩下的是9個屬性(field3-11),每個屬性用數(shù)字表示1000025,2,5,1,1,1,2,1,3,1,1NOTE:16pointswithmissingattributes(indicatedbya0)有16個遺失的屬性,會造成統(tǒng)計不準(zhǔn)確,用0表示classifier分類器分類器是一種計算機程序。他的設(shè)計目標(biāo)是在通過學(xué)習(xí)后,
8、可自動將數(shù)據(jù)分到已知類別。應(yīng)用在搜索引擎以及各種檢索程序中。同時也大量應(yīng)于數(shù)據(jù)分析與預(yù)測領(lǐng)域。分類器是一種機器學(xué)習(xí)程序,因此歸為人工智能的范疇中。人工智能的多個領(lǐng)域,包括數(shù)據(jù)挖掘,專家系統(tǒng),模式識別都用到此類程序。對于分類器,其實質(zhì)為