Modelagem e análise de dados experimentais com o programa computacional R

Disciplina: Tópicos Especiais (AF 722) - Modelagem e análise de dados experimentais com o programa computacional R
Professor Coordenador: PhD. Louise Larissa May De Mio
Professor Coordenador: PhD. Paulo Justiniano Ribeiro Júnior, LEG/UFPR
Professor Colaborador: MSc. Walmes Marques Zeviani, LEG/UFPR
Nº de Créditos: 4
Carga horária: 60h
Período Letivo: 2º semestre de 2012

<bibtex>

@TechReport{ribeiro_jr-leg,

url = {http://www.leg.ufpr.br/~paulojus/embrapa/Rembrapa/Rembrapa.pdf},
year = {2009},
author = {Paulo Justiniano {Ribeiro Júnior}},
title = {Introdução ao Ambiente Estatístico R}

}

@TechReport{faraway-cran,

url = {http://cran.r-project.org/doc/contrib/Faraway-PRA.pdf},
year = {2002},
author = {Julian J. Faraway},
title = {Practical Regression and Anova using {R}}

}

@book{faraway_linear_2005,

      title = {Linear models with R},
      url = {http://www.leg.ufpr.br/~walmes/docs/Linear%20Models%20with%20R.pdf},
isbn = {9781584884255},
publisher = {Chapman \& Hall/{CRC}},
author = {Julian James Faraway},
year = {2005},
pages = {244}

}

@book{venables_modern_2002,

title = {Modern applied statistics with S},
      url = {http://www.leg.ufpr.br/~walmes/docs/Modern%20Applied%20Statistical%20Methods%20with%20S%20plus.pdf},
isbn = {9780387954578},
publisher = {Birkhäuser},
author = {William N. Venables and Brian D. Ripley},
year = {2002},
pages = {518}

}

@book{everitt_handbook_2009,

edition = {2},
title = {A Handbook of Statistical Analyses Using R, Second Edition},
      %url = {http://www.maths.bath.ac.uk/~jjf23/ELM/},
      url = {http://www.leg.ufpr.br/~walmes/docs/A%20handbook%20of%20statistical%20analyses%20using%20R.pdf},
isbn = {1420079336},
publisher = {Chapman \& Hall},
author = {Brian S. Everitt and Torsten Hothorn},
month = jul,
year = {2009}

}

@book{dalgaard_introductory_2008,

edition = {2nd},
title = {Introductory Statistics with R},
      url = {http://www.leg.ufpr.br/~walmes/docs/Introductory%20Statistics%20with%20R.pdf},
isbn = {0387790535},
publisher = {Springer},
author = {Peter Dalgaard},
month = aug,
year = {2008}

}

@book{dobson_introduction_2008,

edition = {3},
title = {An Introduction to Generalized Linear Models, Third Edition},
      url = {http://www.leg.ufpr.br/~walmes/docs/An%20introduction%20to%20generalized%20linear%20models.pdf},
isbn = {1584889500},
publisher = {Chapman and {Hall/CRC}},
author = {Annette J. Dobson and Adrian Barnett},
month = may,
year = {2008}

}

@book{ritz_nonlinear_2008,

edition = {1},
title = {Nonlinear Regression with R},
      url = {http://www.leg.ufpr.br/~walmes/docs/Nonlinear%20Regression%20with%20R.pdf},
isbn = {0387096159},
publisher = {Springer},
author = {Christian Ritz and Jens Carl Streibig},
month = nov,
year = {2008}

}

@book{sarkar_lattice:_2008,

edition = {1},
title = {Lattice: Multivariate Data Visualization with R},
      url = {http://www.leg.ufpr.br/~walmes/docs/Lattice%20Multivariate%20Data%20Visualization%20with%20R.pdf},
isbn = {0387759689},
shorttitle = {Lattice},
publisher = {Springer},
author = {Deepayan Sarkar},
month = mar,
year = {2008}

}

@book{murrell_r_2005,

edition = {1},
title = {R Graphics},
      url = {http://www.leg.ufpr.br/~walmes/docs/R%20Graphics.pdf},
isbn = {{158488486X}},
publisher = {Chapman and {Hall/CRC}},
author = {Paul Murrell},
month = jul,
year = {2005}

}

</bibtex>

Diretório com arquivos de dados e scripts do curso: af722_2012

• Manhã (M): 09:00-10:30, 10:50-12:00;
• Tarde (T): 13:30-15:00, 15:20-14:30;

• Least Squares Regression Line Notes;
• Least Squares Regression;
• Introductory Statistics - Chapter 10: Regression;
• Hypotesis test with R;

#------------------------------------------------------------------------------------------

vol <- read.table("http://www.leg.ufpr.br/~walmes/data/volume.txt",
                  header=TRUE, sep="\t")
str(vol)
vol$dos <- factor(vol$dose)

xyplot(volu~dose|gen, data=vol)

m0 <- aov(volu~gen+dos+gen:dos, data=vol)
anova(m0)

par(mfrow=c(2,2)); plot(m0); layout(1)

boxcox(m0)

m1 <- aov((volu^(1/3))~gen+dos+gen:dos, data=vol)
par(mfrow=c(2,2)); plot(m1); layout(1)
anova(m1)

with(vol, fat2.crd(gen, dos, volu^(1/3), mcomp=c("sk","tukey")))

#------------------------------------------------------------------------------------------

plot(residuals(m0)~vol$dos)
plot(residuals(m0)~vol$dose)

qqmath(~residuals(m0)|vol$dose)

pesos <- tapply(residuals(m0), vol$dose, var)
vol$pesos <- rep(pesos, each=27)

m2 <- aov(volu~gen+dos+gen:dos, data=vol, weights=1/vol$pesos)
par(mfrow=c(2,2)); plot(m2); layout(1)
anova(m2)

require(doBy)

popMeans(m2, effect="gen")
popMeans(m2, effect="dos")
popMeans(m2, effect=c("gen", "dos"))

require(agricolae)
glr <- df.residual(m2)
s2 <- deviance(m2)/df.residual(m2)

with(subset(vol, dose=="0"),
     HSD.test(volu, gen, DFerror=glr, MSerror=pesos[1]*s2))

with(subset(vol, dose=="5"),
     HSD.test(volu, gen, DFerror=glr, MSerror=pesos[2]*s2))

with(subset(vol, dose=="25"),
     HSD.test(volu, gen, DFerror=glr, MSerror=pesos[3]*s2))

#------------------------------------------------------------------------------------------
# dose em cada genótipo

X <- popMatrix(m2, effect=c("gen", "dos"))
contr <- expand.grid(gen=levels(vol$gen), dos=levels(vol$dos))
which(contr$gen=="ATF06B")

contr.x <- rbind("0vs5"=X[1,]-X[10,],
                 "0vs25"=X[1,]-X[19,],
                 "5vs25"=X[10,]-X[19,])
contr.x%*%coef(m2) # estimativas dos contrastes
contr.x%*%vcov(m2)%*%t(contr.x)

summary(glht(m2, linfct=contr.x))

#------------------------------------------------------------------------------------------
# fizemos isso para um único nível de gen, o código abaixo faz para todos

lM <- lapply(levels(vol$gen),
             function(g){
               X[contr$gen==g,]
             })
lM

com <- combn(3, 2)

compr <- lapply(lM,
                function(i){
                  m <- t(apply(com, 2, function(j) i[j[1],]-i[j[2],]))
                })
names(compr) <- levels(vol$gen)
compr

lapply(compr, function(g) summary(glht(m2, linfct=g)))

#------------------------------------------------------------------------------------------
# colocando os resultados em um gráfico com IC

IC <- lapply(compr, function(g) confint(glht(m2, linfct=g)))
IC <- lapply(IC, "[[", "confint") #IC[[1]]$confint
IC <- do.call(rbind, IC)
nm <- apply(com, 2, function(i) paste(levels(vol$dos)[i[1]], levels(vol$dos)[i[2]], sep="vs"))

IC <- cbind(expand.grid(compr=nm, gen=levels(vol$gen)), IC)
str(IC)

require(latticeExtra)

segplot(compr~lwr+upr|gen, data=IC)

segplot(compr~lwr+upr|gen, data=IC, layout=c(1,9),
        strip.left=TRUE, strip=FALSE,
        draw.bands=FALSE, centers=Estimate,
        segments.fun=panel.arrows, ends="both",
        angle=90, length=1, unit="mm")

segplot(compr~lwr+upr|gen, data=IC, layout=c(1,9),
        strip.left=TRUE, strip=FALSE,
        draw.bands=FALSE, centers=Estimate,
        segments.fun=panel.arrows, ends="both",
        angle=90, length=1, unit="mm",
        panel=function(...){
          panel.segplot(...)
          panel.abline(v=0, col=2)
        })

#------------------------------------------------------------------------------------------