實(shí)驗(yàn)報(bào)告 支持向量機(jī)

#實(shí)驗(yàn)報(bào)告支持向量機(jī)實(shí)驗(yàn)原理：支持向量機(jī)的原理和實(shí)現(xiàn)技術(shù)。。實(shí)驗(yàn)題目：對(duì)鳶尾花數(shù)據(jù)利用SVM技術(shù)進(jìn)行分類(lèi)預(yù)測(cè)。實(shí)驗(yàn)要求：把鳶尾花數(shù)據(jù)分成訓(xùn)練集和測(cè)試集，然后針對(duì)樣本SVM預(yù)測(cè)分類(lèi)實(shí)驗(yàn)題目--分析報(bào)告：data(iris)>rm(list=ls())>gc()used(Mb)gctrigger(Mb)maxused(Mb)Ncells25214313.560839432.541051522.0Vcells5284864.1838860864.0160673612.3>library(MASS)>library(MASS)>data(iris)>library(e1071)>summary(iris)Sepal.LengthSepal.WidthMin.:4.300Min.:2.0001stQu.:5.1001stQu.:2.800Median:5.800Median:3.000Mean:5.843Mean:3.0573rdQu.:6.4003rdQu.:3.300Max.:7.900Max.:4.400Petal.LengthPetal.WidthMin.:1.000Min.:0.1001stQu.:1.6001stQu.:0.300Median:4.350Median:1.300Mean:3.758Mean:1.1993rdQu.:5.1003rdQu.:1.800Max.:6.900Max.:2.500Speciessetosa:50versicolor:50virginica:50僅選擇Petal.Length和Petal.Width這兩個(gè)特征時(shí)>model<-svm(Petal.Length~Petal.Width,data=iris)>print(model)Call:svm(formula=Petal.Length~Petal.Width,data=iris)Parameters:SVM-Type:eps-regressionSVM-Kernel:radialcost:1gamma:1epsilon:0.1NumberofSupportVectors:80>summary(model)Call:svm(formula=Petal.Length~Petal.Width,data=iris)Parameters:SVM-Type:eps-regressionSVM-Kernel:radialcost:1gamma:1epsilon:0.1NumberofSupportVectors:80>svm(formula=Petal.Length~Petal.Width,data=iris)Call:svm(formula=Petal.Length~Petal.Width,data=iris)Parameters:SVM-Type:eps-regressionSVM-Kernel:radialcost:1gamma:1epsilon:0.1NumberofSupportVectors:80>predict(model,iris)12312341.4238371.4238371.4238371.42383791011121.4238371.4569131.4238371.423837171819201.6164691.4770961.4770961.477096252627281.4238371.4238371.6164691.423837333435361.4569131.4238371.4238371.423837414243441.4770961.4770961.4238372.115572495051521.4238371.4238374.4971394.675747575859604.8369863.4771784.2931204.497139656667684.2931204.4971394.6757473.47717856781.4238371.6164691.4770961.423837131415161.4569131.4569131.4238371.616469212223241.4238371.6164691.4238371.834129293031321.4238371.4238371.4238371.616469373839401.4238371.4569131.4238371.423837454647481.6164691.4770961.4238371.423837535455564.6757474.2931204.6757474.293120616263643.4771784.6757473.4771784.497139697071724.6757473.7845145.1343674.29312073747576777879804.6757474.0568294.2931204.4971394.4971394.9881834.6757473.47717881828384858687883.7845143.4771784.0568294.8369864.6757474.8369864.6757474.29312089909192939495964.2931204.2931204.0568294.4971394.0568293.4771784.2931204.0568299798991001011021031044.2931204.2931203.7845144.2931205.7523445.2771775.5403125.1343671051061071081091101111125.6463545.5403124.9881835.1343675.1343675.7523445.4144235.2771771131141151161171181191205.5403125.4144235.7604485.7228085.1343675.6463545.7228084.6757471211221231241251261271285.7228085.4144235.4144235.1343675.5403125.1343675.1343675.1343671291301311321331341351365.5403124.8369865.2771775.4144235.6463544.6757474.4971395.7228081371381391401411421431445.7604485.1343675.1343675.5403125.7604485.7228085.2771775.7228081451461471481491505.7523445.7228085.2771775.4144235.7228085.134367分割數(shù)據(jù)集>set.seed(2)>test=sample(1:nrow(iris),100)>iris.train<-iris[-test,]>iris.test<-iris[test,]>dim(iris.train);dim(iris.test)[1]505[1]1005

>model<-svm(Petal.Length~Petal.Width,data=iris.train)prediction<-predict(model,iris.test[,-1])tab<-table(predtab<-table(pred=prediction,trueris.test[,1])tabtrueed4.34.44.64.84.955.1525.4555.6575.85961.476426185233510122142221001001.514934525019011001200100000001.537794306199520001012000010001.696750649553480000011030000001.940503187907960000001000000002.247766796584170000010000000003.593325051594760000110001010013.857169576777270000001001100004.075688311422150000000001011004.254279229338620000000000220004.404109854107820000000100000004.539639007089480000000010100124.675688856047920000000000000014.824398220950480000100000000004.992496243428280000000000000005.17942227280050000000000002005.376786409977140000000000100005.569454200330490000000000000005.738170283834570000000000000005.863218184264910000000000000005.923719404515560000000000000005.92829837440344000000000000100truepred6.16.26.36.46.56.66.76.86.97.27.37.47.77.91.47642618523351000000000000001.51493452501901000000000000001.53779430619952000000000000001.69675064955348000000000000001.94050318790796000000000000002.24776679658417000000000000003.59332505159476000000000000003.85716957677727000000000000004.07568831142215100000000000004.25427922933862001101000000004.40410985410782300001000000004.53963900708948001100100000004.67568885604792001000000100004.82439822095048000000100000004.99249624342828110110100110005.1794222728005001000000001005.37678640997714000000000000015.56945420033049000000011000005.73817028383457000010000000105.86321818426491010100012000105.92371940451556001000100000005.9282983744034400100010000000>classAgreement(tab)$'diag'[1]0.02$kappa[1]-0.01554404$rand[1]0.910101$crand[1]0.05377635>tuned<-tune.svm(Petal.Length~Petal.Width,data=iris.train,gamma=10八(-6:-1),+cost=10八(1:2))>summary(tuned)Parametertuningof‘svm':samplingmethod:10-foldcrossvalidationbestparameters:gammacost0.110bestperformance:0.1490541-Detailedperformanceresults:gammacosterrordispersion11e-06103.47224882.206454121e-05103.42808442.180270331e-0410