mba論文評(píng)估幾種流行學(xué)習(xí)降維分類器應(yīng)用于癌癥數(shù)據(jù)的性能pdf

1、分 類 編 號(hào) :密 級(jí) :單 位 代 碼 ： 100 65號(hào) ： 092 0900 6研 究 生 學(xué)位 論 文論文題 目：評(píng)估 幾種 流行 學(xué) 習(xí) 降維分類器 應(yīng)用 于 癌 癥 數(shù) 據(jù) 的 性 能學(xué) 生 姓 名 ：高麗 申請(qǐng) 學(xué)位級(jí)別 ： 碩 士申請(qǐng)專業(yè)名稱 ： 孝 女 tf技術(shù)學(xué)研 究 方 向： 數(shù)據(jù)挖掘指導(dǎo)教師姓名 ：王淑琴 專業(yè)技術(shù)職稱 ： 副教授提 交 論 文 日期 ：20 12 年 4 月 天 洋 師 范 人

2、學(xué)碩 士 學(xué)位 論文A B S T R A C TT he l arge num ber of hi gh-di m ensi onalcancer dat a has br oughtus a l otof t roubl e.H i gh-di m ensi onalcancer data have a si gni fi cantf eat ure t hati s di m ensi on m ore t ha

3、nsam pl e.In t he face of such hi gh-di m ensi onaldata,dat a di m ensi on reduct i on hasbecom e an essent i alst ep i n dat a m i ni ng.D i m ensi on reduct i on techni ques can reducet he di m ensi onalst ruct ure a

4、nd keep l ocalcharact eri st ics t hrough ext ractfeat ure,tohel p m achi ne l ear ni ng m et hod for accurat e effi ci entcl assi f i cat i on.H ow ever,t hatch o o se w h at k in d o f d im e n sio n re d u ctio n c

5、 la ssifi er to a ch iev e th e b e st e ff e ct o fclassifi cation ,is th e cen tralcon ten t o f t h is t h esis to research .T h is t h esis discu ssesei ghtdi m ensi on-reduct i on m et hods,w hich are based on

6、t he m ani fol d struct ure.L i neardi m ensi on-reduction m ethods are P ri nci palC om ponentA nal ysis (P C A ),F isher L i neardi scri m i n antanal ysis (L D A ),M ul ti -di m ensi onalScal i ng (M D S),L ocali

7、 ty Preservi ngProject i on (L PP).N on-L i near di m ensi on-reduct i on m et hods are L ocalL i nearE m bedding (L L E ),L aplacian E igenm ap (L E )， Isom et ric m appi ng (IS O M A P ),C l ass-w i se N on-L ocal i t

8、y Preserving P rojecti on (C N L P P).L P P i s al so a m ani fol d l earni ngm et hod,notonl y keep t he dat a setof nonl i near charact eri st i cs,and w i t h l i near m et hodcharact eri sti cs of si m pl e

9、cal cul ati on,di rectqui ck .L P P al gori t hm of di m ensi onreduct i on i n cancer d at a sho w ed t he adv an t age.T h e m a in c o n te n t o f th e fu llte x t su m m ariz e d a s fo llo w s :(1)B ri

10、efl y sum m ari zed i n t he fi el d of hi gh di m ensi onaldat a processi ng stat us andcont ai ns som e problem s.A nd do som e anal ysis for t he t ypi caldi m ensi on reducti onm et hods,i ncl udi ng al gori

11、thm ,advantages,disadvantages,and the presentresearchst at us.Item phati cal l y i nt roduces t he L P P.(2) T hi s t hesi s i nt roduced t he basi c pri nci pl e of supportvector m achi ne and t hegeneralcl assi fi

12、cati on process.H ow to use M atl ab 2009 t o real i ze supportvect orm achi nes (SV M )al gori t hm for t he cl assi ficati on of cancer data.(3)T hi s t hesi s st udied thatt he perform ance of cl assi fi es w h

13、i ch are com bi ned byS V M ,N N C and som e d i m ensi on reducti on t echni ques i n cancer dat a cl assi fi cat i onm odel ,w hi ch provi de t he reference for t he cl assifi cat i on of rel at ed.T hi s t hesi s us