路徑模型和PLS課件

路徑模型和PLS吳喜之路徑模型和PLS吳喜之1基于回歸的傳統(tǒng)方法的假定

(e.g.,multipleregressionanalysis,discriminantanalysis,logisticregression,analysisofvariance)簡(jiǎn)單模型結(jié)構(gòu):Thepostulationofasimplemodelstructure(atleastinthecaseofregression-basedapproaches);變量是可觀測(cè)的:Theassumptionthatallvariablescanbeconsideredasobservable;所有變量可精確測(cè)量:Theconjecturethatallvariablesaremeasuredwithouterror,whichmaylimittheirapplicabilityinsomeresearchsituations.基于回歸的傳統(tǒng)方法的假定

(e.g.,multipler2為克服第一代基于回歸的模型的弱點(diǎn)

Structuralequationmodeling(SEM)SEM僅同時(shí)分析自變量和因變量之間的鏈接中的一層.SEM允許多個(gè)自變量和因變量結(jié)構(gòu)中的關(guān)系的同時(shí)建模.因此不再區(qū)別因變量和自變量,但是區(qū)別外生和內(nèi)生隱變量變量(theexogenousandendogenouslatentvariables),前者不被設(shè)定的模型所解釋(總是因變量),后者為被解釋變量.SEM能夠構(gòu)造由指標(biāo)變量(indicators,items,manifestvariables,orobservedmeasures)以及可觀測(cè)變量的度量誤差來度量的不可觀測(cè)變量為克服第一代基于回歸的模型的弱點(diǎn)

Structurale3兩種模型基于協(xié)方差(或最大似然)的方法:Covariance-basedSEM(軟件工具:EQS,AMOS,SEPATH,andCOSAN,theLISREL)基于方差(成分)的方法:Variance-basedSEM(Component-basedSEM),andtopresentpartialleastsquares(PLS)兩種模型基于協(xié)方差(或最大似然)的方法:Covarianc4內(nèi)生和外生隱變量的關(guān)系內(nèi)生隱變量及其指標(biāo)及測(cè)量誤差的關(guān)系外生隱變量及其指標(biāo)及測(cè)量誤差的關(guān)系內(nèi)生和外生隱變量的關(guān)系內(nèi)生隱變量及其指標(biāo)及測(cè)量誤差的關(guān)系外生5名詞η(eta)=latentendogenousvariable;ξ(xi)=latentexogenous(i.e.,independent)variable;ζ(zeta)=randomdisturbanceterm;“errorsinequations”γ(gamma)=pathcoefficient;φ(phi)noncausalrelationshipbetweentwolatentexogenousvariables;yi=indicatorsofendogenousvariables;εi(epsilon)=measurementerrorsforindicatorsofendogenousvariable;λyi(lambday)=loadingsofindicatorsofendogenousvariable;xi=indicatorsofendogenousvariable;δi(delta)=measurmenterrorsforindicatorsofexogenousvariable;λxi=(lambdax)loadingsofindicatorsofexogenousvariable.名詞η(eta)=latentendogenous6內(nèi)生和外生隱變量的關(guān)系:theoreticalequations:representingnonobservationalhypothesesandtheoreticaldefinitions(structuralmodel)內(nèi)生隱變量及其指標(biāo)及測(cè)量誤差的關(guān)系(measurementequations)(measurementmodel)外生隱變量及其指標(biāo)及測(cè)量誤差的關(guān)系(measurementequations)(measurementmodel)矩陣記號(hào)結(jié)構(gòu)模型度量模型內(nèi)生和外生隱變量的關(guān)系:theoreticalequat7三種不同類型的不可觀測(cè)變量原則上不可觀測(cè)變量:variablesthatareunobservableinprinciple(e.g.,theoreticalterms);

原則上不可觀測(cè),但暗含經(jīng)驗(yàn)概念或能夠從觀測(cè)值導(dǎo)出:variablesthatareunobservableinprinciplebuteitherimplyempiricalconceptsorcanbeinferredfromobservations(e.g.,attitudes,whichmightbereflectedinevaluations);用可觀測(cè)變量定義的不可觀測(cè)變量:unobservablevariablesthataredefinedintermsofobservables.三種不同類型的不可觀測(cè)變量原則上不可觀測(cè)變量:variab8兩類指標(biāo)變量:a)reflectiveindicatorsthatdependontheconstruct;b)formativeones(alsoknownascausemeasures)thatcausetheformationoforchangesinanunobservablevariable兩類指標(biāo)變量:a)reflectiveindicato9二者的區(qū)別Reflectiveindicatorsshouldhaveahighcorrelation(astheyarealldependentonthesameunobservablevariable),formativeindicatorsofthesameconstructcanhavepositive,negative,orzerocorrelationwithoneanother(Hulland,1999),whichmeansthatachangeinoneindicatordoesnotnecessarilyimplyasimilardirectionalchangeinothers(Chin,1998a).二者的區(qū)別Reflectiveindicatorssho10基于協(xié)方差(SEM-ML)和基于方差(SEM-PLS)的兩種建?；趨f(xié)方差方法試圖減少樣本協(xié)方差和理論預(yù)測(cè)的協(xié)方差的區(qū)別,因此參數(shù)估計(jì)過程試圖重新產(chǎn)生觀測(cè)到協(xié)方差矩陣(先計(jì)算模型參數(shù),然后用回歸得到個(gè)體估計(jì)值)基于方差的方法:使得被自變量解釋的因變量方差最大,而不是再生經(jīng)驗(yàn)協(xié)方差矩陣.除了結(jié)構(gòu)模型和測(cè)量模型之外,PLS有第三部分:用來估計(jì)隱變量的個(gè)體值的加權(quán)關(guān)系(weightrelations)(先計(jì)算個(gè)體值—不可觀測(cè)變量值用他們的指標(biāo)變量的線性組合表示,所用權(quán)重使得最終的個(gè)體值反映了因變量的大多數(shù)方差,再估計(jì)不可觀測(cè)變量的估計(jì)值.最后確定結(jié)構(gòu)模型的參數(shù).)基于協(xié)方差(SEM-ML)和基于方差(SEM-PLS)的兩種11PLS估計(jì)步驟:兩步確定權(quán)重(wi):第一步:外部近似(類似于主成份分析forreflective,回歸forformativeindicators)

第二步:內(nèi)部近似(三種方法:centroid,factor,andpathweightingscheme)

得到更新的重復(fù)這兩步直到收斂PLS估計(jì)步驟:兩步確定權(quán)重(wi):得到更新的重復(fù)這兩步12 PLS優(yōu)點(diǎn):沒有總體假定或度量標(biāo)度的假定,因此也沒有分布假定.然而需要某些假定,如線性回歸的系統(tǒng)部分等于因變量的條件期望.根據(jù)MonteCarlo模擬,PLS非常穩(wěn)健,而且隱變量的得分總是和真值吻合.

由于隱變量的個(gè)體值為顯變量的整合,由于后者的度量誤差,該值為不相合的(但漸近相合).由于樣本及每個(gè)隱變量的指標(biāo)的有限性,PLS有低估隱變量之間的相關(guān)及高估載荷(測(cè)量變量的系數(shù))的傾向. PLS優(yōu)點(diǎn):沒有總體假定或度量標(biāo)度的假定,因此也沒有13在基于協(xié)方差和基于方差的SEM之間的選擇在每個(gè)隱變量的指標(biāo)變量數(shù)目太大時(shí),基于協(xié)方差的SEM就沒有辦法了.而實(shí)際上,如果沒有足夠的指標(biāo)變量(有時(shí)達(dá)到500個(gè)),不能做任何嚴(yán)肅的路徑模型研究.由于有充分多的指標(biāo)變量,選擇權(quán)重不會(huì)對(duì)路徑系數(shù)有任何影響,相合性問題就不是問題了.Therefore,theresearcherwouldbewelladvisedtousePLSinsteadofcovariance-basedSEMinsuchsituations.RecapitulatingtheseargumentsbyusingthewordsofS.Wold(1993),H.Wold’sson,onecansaythat“thenaturaldomainforLV[latentvariable]modelssuchasPLS…iswherethenumberof‘significant’LV’sissmall,muchsmallerthanthenumberofmeasuredvariables…andthanthenumberofobservations.”(p.137).在基于協(xié)方差和基于方差的SEM之間的選擇在每個(gè)隱變量的指標(biāo)變14其它PLS占優(yōu)勢(shì)的情況Constructsaremeasuredprimarilybyformativeindicators.那時(shí)基于協(xié)方差的方法(LISREL)會(huì)有嚴(yán)重的識(shí)別困難LISREL至少要100,甚至200個(gè)觀測(cè)值,但PLS只需50(甚至在兩個(gè)隱變量,27個(gè)顯變量時(shí)只有10個(gè)觀測(cè)值的情況).其它PLS占優(yōu)勢(shì)的情況Constructsaremeas15Sohn&Park(2001)[3]的蒙特卡羅模擬比較表明：（1）以均方誤差和對(duì)因子載荷的方差為標(biāo)準(zhǔn)，在數(shù)據(jù)量小，而且表現(xiàn)出稍微非正態(tài)時(shí)，ML性能最差；當(dāng)數(shù)據(jù)是正態(tài)或近似正態(tài)時(shí)，在ML和PLS之間沒有顯著差別，（2）以因子載荷的偏差為標(biāo)準(zhǔn)，無論數(shù)據(jù)量大小，ML隨著非正態(tài)增加而性能變差，（3）以回歸系數(shù)的均方誤差為標(biāo)準(zhǔn)，PLS比ML要好。

Sohn&Park(2001)[3]的蒙特卡羅模擬比較表16顧客滿意度模型顧客滿意度模型17瑞典顧客滿意度指數(shù)模型感知表現(xiàn)顧客預(yù)期質(zhì)量顧客滿意度顧客抱怨顧客忠誠(chéng)SCSB感知表現(xiàn)顧客預(yù)期質(zhì)量顧客滿意度顧客抱怨顧客忠誠(chéng)五個(gè)隱含變量中，顧客預(yù)期質(zhì)量為外生隱變量(exogenouslatentvariable)，其余為內(nèi)生隱變量(endogenouslatentvariable)。瑞典顧客滿意度指數(shù)模型感知表現(xiàn)顧客預(yù)期質(zhì)量顧客滿意度顧客抱怨18感知質(zhì)量軟件預(yù)期質(zhì)量顧客滿意度顧客忠誠(chéng)感知價(jià)值感知質(zhì)量硬件形象ECSI歐洲顧客滿意度指數(shù)模型感知質(zhì)量軟件感知質(zhì)量硬件感知價(jià)值預(yù)期質(zhì)量形象顧客滿意度顧客忠誠(chéng)感知質(zhì)量軟件預(yù)期質(zhì)量顧客滿意度顧客忠誠(chéng)感知價(jià)值感知質(zhì)量硬件形19感知質(zhì)量（可分為產(chǎn)品和服務(wù)兩部分）預(yù)期質(zhì)量顧客滿意度(ACSI)顧客抱怨顧客忠誠(chéng)度感知價(jià)值A(chǔ)CSI美國(guó)顧客滿意度指數(shù)模型感知質(zhì)量感知價(jià)值預(yù)期質(zhì)量顧客滿意度顧客抱怨顧客忠誠(chéng)度感知質(zhì)量預(yù)期質(zhì)量顧客滿意度(ACSI)顧客抱怨顧客忠誠(chéng)度感知20感知質(zhì)量（可分為產(chǎn)品和服務(wù)兩部分）預(yù)期質(zhì)量顧客滿意度(ACSI)顧客抱怨顧客忠誠(chéng)度感知價(jià)值A(chǔ)CSI滿足顧客需求程度整體印象滿足顧客需求程度可靠性可靠性整體印象質(zhì)量?jī)r(jià)格比未確認(rèn)期望值與理想之距離總體滿意度向經(jīng)理抱怨向雇員抱怨再購(gòu)可能性價(jià)格承受度價(jià)格質(zhì)量比美國(guó)顧客滿意度指數(shù)模型感知質(zhì)量預(yù)期質(zhì)量顧客滿意度(ACSI)顧客抱怨顧客忠誠(chéng)度感知21感知質(zhì)量h2預(yù)期質(zhì)量h1顧客滿意度h4顧客忠誠(chéng)度h5感知價(jià)值h3品牌形象h6中國(guó)耐用消費(fèi)品滿意度指數(shù)框圖總體感知質(zhì)量x5自定義感知質(zhì)量x6可靠性感知質(zhì)量x7服務(wù)感知質(zhì)量x8可靠性預(yù)期質(zhì)量x3品牌總體印象x17品牌特征顯著度x18價(jià)格質(zhì)量比x9再購(gòu)可能性x15與理想之距離x14總體滿意度x11與其他品牌距離x13與期望之距離x12質(zhì)量?jī)r(jià)格比x10價(jià)格承受度x16總體預(yù)期質(zhì)量x1自定義預(yù)期質(zhì)量x2服務(wù)預(yù)期x4中國(guó)耐用消費(fèi)品顧客滿意度指數(shù)模型感知質(zhì)量h2預(yù)期質(zhì)量h1顧客滿意度h4顧客忠誠(chéng)度h5感知價(jià)值22感知質(zhì)量顧客滿意度顧客忠誠(chéng)感知價(jià)值品牌形象中國(guó)非耐用消費(fèi)品顧客滿意度指數(shù)框圖總體感知質(zhì)量感知質(zhì)量指標(biāo)1感知質(zhì)量指標(biāo)2感知質(zhì)量指標(biāo)n品牌總體印象品牌特征顯著度價(jià)格質(zhì)量比再購(gòu)可能性與理想之距離總體滿意度與其他品牌距離質(zhì)量?jī)r(jià)格比價(jià)格承受度中國(guó)非耐用消費(fèi)品顧客滿意度指數(shù)模型感知質(zhì)量顧客滿意度顧客忠誠(chéng)感知價(jià)值品牌形象中國(guó)非耐用消費(fèi)品顧23感知質(zhì)量預(yù)期質(zhì)量顧客滿意度顧客忠誠(chéng)感知價(jià)值品牌形象中國(guó)服務(wù)行業(yè)顧客滿意度指數(shù)框圖總體感知質(zhì)量響應(yīng)性感知質(zhì)量可靠性感知質(zhì)量保證性感知質(zhì)量移情性感知質(zhì)量有形性感知質(zhì)量總體預(yù)期質(zhì)量品牌總體印象品牌特征顯著度價(jià)格質(zhì)量比回頭可能性與理想之距離總體滿意度與其他品牌距離與期望之距離質(zhì)量?jī)r(jià)格比價(jià)格承受度中國(guó)服務(wù)行業(yè)顧客滿意度指數(shù)模型感知質(zhì)量預(yù)期質(zhì)量顧客滿意度顧客忠誠(chéng)感知價(jià)值品牌形象中國(guó)服務(wù)行24感知質(zhì)量h2預(yù)期質(zhì)量h1顧客滿意度h4顧客忠誠(chéng)度h5感知價(jià)值h3品牌形象h6中國(guó)耐用消費(fèi)品滿意度指數(shù)框圖總體感知質(zhì)量x5自定義感知質(zhì)量x6可靠性感知質(zhì)量x7服務(wù)感知質(zhì)量x8可靠性期質(zhì)量x3品牌總體印象x17品牌特征顯著度x18價(jià)格質(zhì)量比x9(Pricegivenquality)再購(gòu)可能性x15與理想之距離x14總體滿意度x11與其他品牌距離x13與期望之距離x12質(zhì)量?jī)r(jià)格比x10(Qualitygivenprice)價(jià)格承受度x16總體預(yù)期質(zhì)量x1自定義預(yù)期質(zhì)量x2服務(wù)預(yù)期x4中國(guó)耐用消費(fèi)品顧客滿意度指數(shù)模型感知質(zhì)量h2預(yù)期質(zhì)量h1顧客滿意度h4顧客忠誠(chéng)度h5感知價(jià)值25這里，包含有b的B矩陣、h及z是未知的。而B矩陣的形式完全被圖模型所確定。這里，包含有b的B矩陣、h及z是未知的。而B矩陣的形式完全被26這里，包含有l(wèi)的L矩陣、h是未知的，而x是可觀測(cè)的。而L矩陣的形式完全被圖模型所確定。這里，包含有l(wèi)的L矩陣、h是未知的，而x是可觀測(cè)的。而L矩陣27偏最小二乘(PLS)法

解

路徑模型(PathModel)吳喜之(plspm)偏最小二乘(PLS)法

解

路徑模型(PathModel)28例子（先不看數(shù)字）例子（先不看數(shù)字）29其中：reflectiveindicators

“l(fā)oadings”其中：reflectiveindicators

“l(fā)oad30其中：reflectiveindicators

“weights”其中：reflectiveindicators

“weig31library(plspm)##typicalexampleofPLS-PMincustomersatisfactionanalysis##modelwithsixLVsandreflectiveindicatorsdata(satisfaction)IMAG<-c(0,0,0,0,0,0)EXPE<-c(1,0,0,0,0,0)QUAL<-c(0,1,0,0,0,0)VAL<-c(0,1,1,0,0,0)SAT<-c(1,1,1,1,0,0)LOY<-c(1,0,0,0,1,0)sat.mat<-rbind(IMAG,EXPE,QUAL,VAL,SAT,LOY)sat.sets<-list(1:5,6:10,11:15,16:19,20:23,24:27)sat.mod<-rep("A",6)##reflectiveindicatorsres2<-plspm(satisfaction,sat.mat,sat.sets,sat.mod,scheme="centroid",scaled=FALSE)##plotdiagramoftheinnermodelplot(res2)##plotdiagramsofboththeinnermodelandoutermodel(loadingsandweights)plot(res2,what="weights")plot(res2,what="loadings")plot(res2,what="all")##End(Notrun)library(plspm)32程序plspm(x,inner.mat,sets,modes=NULL,scheme="centroid",scaled=TRUE,boot.val=FALSE,br=NULL,plsr=FALSE)xAnumericmatrixordataframecontainingthemanifestvariables.inner.matAsquare(lowertriangular)booleanmatrixindicatingthepathrelationshipsbetwennlatentvariables.setsListofvectorswithcolumnindicesfromxindicatingwhichmanifestvariablescorrespondtothelatentvariables.modesAcharactervectorindicatingthetypeofmeasurementforeachlatentvariable."A"forreflectivemeasurementor"B"forformativemeasurement(NULLbydefault).schemeAstringofcharactersindicatingthetypeofinnerweightingscheme.Possiblevaluesare"centroid"or"factor".scaledAlogicalvalueindicatingwhetherscalingdataisperformed(TRUEbydefault).boot.valAlogicalvalueindicatingwhetherbootstrapvalidationisperformed(FALSEbydefault).brAnintegerindicatingthenumberbootstrapresamples.Usedonlywhenboot.val=TRUE.plsrAlogicalvalueindicatingwhetherplsregressionisapplied(FALSEbydefault).程序plspm(x,inner.mat,sets,mo33輸出outer.modResultsoftheouter(measurement)model.Includes:outerweights,standardizedloadings,communalities,andredundancies.inner.modResultsoftheinner(structural)model.Includes:pathcoefficientsandR-squaredforeachendogenouslatentvariable.latentsMatrixofstandardizedlatentvariables(variance=1calculateddividedbyN)obtainedfromcentereddata(mean=0).scoresMatrixoflatentvariablesusedtoestimatetheinnermodel.Ifscaled=FALSEthenscoresarelatentvariablescalculatedwiththeoriginaldata(non-stardardized).Ifscaled=TRUEthenscoresandlatentshavethesamevalues.out.weightsVectorofouterweights.loadingsVectorofstandardizedloadings(i.e.correlationswithLVs.)path.coefsMatrixofpathcoefficients(thismatrixhasasimilarformasinner.mat).r.sqrVectorofR-squaredcoefficients.Anobjectofclass"plspm".Whenthefunctionplspm.fitiscalled,itreturnsalistwithbasicresults:輸出outer.modResultsoftheoute34輸出outer.corCorrelationsbetweenthelatentvariablesandthemanifestvariables(alsocalledcrossloadings).inner.sumSummarizedresultsbylatentvariableoftheinnermodel.Includes:typeofLV,typeofmeasurement,numberofindicators,R-squared,averagecommunality,averageredundancy,andaveragevarianceextractedeffectsPatheffectsofthestructuralrelationships.Includes:direct,indirect,andtotaleffects.unidimResultsforcheckingtheunidimensionalityofblocks(Theseresultsareonlymeaningfulforreflectiveblocks).gofTablewithindexesofGoodness-of-Fit.Includes:absoluteGoF,relativeGoF,outermodelGoF,andinnermodelGoF.dataDatamatrixcontainingthemanifestvariablesusedinthemodel.bootListofbootstrappingresults;onlyavailablewhenargumentboot.val=TRUE.Ifthefunctionplspmiscalled,thepreviouslistofresultsalsocontainsthefollowingelements:輸出outer.corCorrelationsbetwee35##typicalexampleofPLS-PMincustomersatisfactionanalysis##modelwithsixLVsandreflectiveindicatorsdata(satisfaction)IMAG<-c(0,0,0,0,0,0)EXPE<-c(1,0,0,0,0,0)QUAL<-c(0,1,0,0,0,0)VAL<-c(0,1,1,0,0,0)SAT<-c(1,1,1,1,0,0)LOY<-c(1,0,0,0,1,0)sat.mat<-rbind(IMAG,EXPE,QUAL,VAL,SAT,LOY)sat.sets<-list(1:5,6:10,11:15,16:19,20:23,24:27)sat.mod<-rep("A",6)##reflectiveindicatorsres2<-plspm(satisfaction,sat.mat,sat.sets,sat.mod,scaled=FALSE)summary(res2)plot(res2)##typicalexampleofPLS-PM36路徑模型和PLSppt課件37res2$unidimres2$unidim38res2$outer.modres2$out.weights輸出第1列res2$loadings輸出第2列res2$outer.modres2$out.weights39路徑模型和PLSppt課件40res2$inner.modres2$path.coefsres2$r.sqrres2$inner.modres2$path.coefsr41res2$inner.sumres2$gofres2$inner.sumres2$gof42res2$latents：輸出所有觀測(cè)值的latent值res2$scores：輸出所有觀測(cè)值的latentscores值res2$latents：輸出所有觀測(cè)值的latent值43res2$effects#即路徑系數(shù)path.coefres2$effects#即路徑系數(shù)path.coef44路徑模型和PLSppt課件45例data(arizona)ari.inner<-matrix(c(0,0,0,0,0,0,1,1,0),3,3,byrow=TRUE)dimnames(ari.inner)<-list(c("ENV","SOIL","DIV"),c("ENV","SOIL","DIV"))ari.outer<-list(c(1,2),c(3,4,5),c(6,7,8))ari.mod<-c("B","B","B")##formativeindicatorsres1<-plspm(arizona,inner=ari.inner,outer=ari.outer,modes=ari.mod,scheme="factor",scaled=TRUE,plsr=TRUE)res1summary(res1)例data(arizona)46plot(res1,what="all")plot(res1,what="all")47例

##exampleofPLS-PMinmulti-blockdataanalysis##estimateapathmodelforthewinedataset##requirespackageFactoMineRlibrary(FactoMineR)data(wine)SMELL<-c(0,0,0,0)VIEW<-c(1,0,0,0)SHAKE<-c(1,1,0,0)TASTE<-c(1,1,1,0)wine.mat<-rbind(SMELL,VIEW,SHAKE,TASTE)wine.sets<-list(3:7,8:10,11:20,21:29)wine.mods<-rep("A",4)#usingfunctionplspm.fit(basicplsalgorithm)res4<-plspm.fit(wine,wine.mat,wine.sets,wine.mods,scheme="centroid")plot(res4,what="all",arr.pos=.4,p=.4,cex.txt=.8)

##End(Notrun)例##exampleofPLS-PMinmult48路徑模型和PLSppt課件49路徑模型和PLSppt課件50路徑模型和PLSppt課件51##Notrun:##examplewithcustomersatisfactionanalysis##groupcomparisonbasedonthesegmentationvariable"gender"data(satisfaction)IMAG<-c(0,0,0,0,0,0)EXPE<-c(1,0,0,0,0,0)QUAL<-c(0,1,0,0,0,0)VAL<-c(0,1,1,0,0,0)SAT<-c(1,1,1,1,0,0)LOY<-c(1,0,0,0,1,0)sat.inner<-rbind(IMAG,EXPE,QUAL,VAL,SAT,LOY)sat.outer<-list(1:5,6:10,11:15,16:19,20:23,24:27)sat.mod<-rep("A",6)##reflectiveindicatorspls<-plspm(satisfaction,sat.inner,sat.outer,sat.mod,scheme="factor",scaled=FALSE)##permutationtestwith100permutationsres.group<-plspm.groups(pls,satisfaction$gender,method="permutation",reps=100)res.groupplot(res.group)

##End(Notrun)plspm.groups{plspm}:GroupComparisoninPLS-PM##Notrun:plspm.groups{plsp52路徑模型和PLSppt課件53路徑模型和PLSppt課件54nipals{plspm}:Non-linearIterativePartialLeastSquares(主成份分析)PrincipalComponentAnalysiswithNIPALSalgorithmlibrary(plspm)data(wines)nip1<-nipals(wines[,-1],nc=5)plot(nip1)nipals{plspm}:Non-linearIte55##USArrestsdatavarynip2<-nipals(USArrests)plot(nip2)##USArrestsdatavary56plsca{plspm}:PLS-CA:PartialLeastSquaresCanonicalAnalysis(典型相關(guān)分析)##exampleofPLSCAwiththevehiclesdatasetdata(vehicles);head(vehicles)names(vehicles)[1]"diesel""turbo""two.doors""hatchback""wheel.base"[6]"length""width""height""curb.weight""eng.size"[11]"horsepower""peak.rpm""price""symbol""city.mpg"[16]"highway.mpg"can<-plsca(vehicles[,1:12],vehicles[,13:16])canplot(can)plsca{plspm}:PLS-CA:Partial57路徑模型和PLSppt課件58路徑模型和PLSppt課件59路徑模型和PLSppt課件60semPLSsemPLS61library(semPLS)#下面是如何構(gòu)建一個(gè)模型(以ECSI為例)#gettingthepathtothe.csvfilerepresentingtheinnerModelptf_Struc<-system.file("ECSIstrucmod.csv",package="semPLS")#gettingthepathtothe.csvfilerepresentingtheouterModelsptf_Meas<-system.file("ECSImeasuremod.csv",package="semPLS")sm<-as.matrix(read.csv(ptf_Struc))(w=read.csv(ptf_Struc))mm<-as.matrix(read.csv(ptf_Meas))構(gòu)建一個(gè)模型(以ECSI為例)library(semPLS)構(gòu)建一個(gè)模型(以ECSI為例)62路徑模型和PLSppt課件63ExpectationQualityValueImageSatisfactionComplaintsLoyaltyECSIExpectationQualityValueImageSa64EC