install.packages("FactoMineR")
install.packages("factoextra")
install.packages("corrplot")

library("FactoMineR")
library("factoextra")

## Read in data and assign variable names
lpga2008 <- read.fwf("http://www.stat.ufl.edu/~winner/data/lpga2008.dat", 
   width=c(30,8,8,8,8,8,8,8,8,8,8), 
   col.names=c("golfer","drive","frwy","grnreg",
      "puttrnd","sandrnd","sandsv","przrnd","lnprzrnd","rounds","golferid"))
attach(lpga2008)
# lpga2008

## Create new variables for analysis that remove outlying golfer(s)
drive1 <- drive[grnreg >= 50]
frwy1 <- frwy[grnreg >= 50]
grnreg1 <- grnreg[grnreg >= 50]
puttrnd1 <- puttrnd[grnreg >= 50]

## Create new data frame with only the 4 variables of interest
lpga1 <- data.frame(drive1,frwy1,grnreg1,puttrnd1)

detach(lpga2008)
attach(lpga1)

X <- as.matrix(lpga1)

res.pca <- PCA(X, graph=FALSE)
print(res.pca)

eig.val <- get_eigenvalue(res.pca)
eig.val

win.graph(width=7.0, height=5.5)
fviz_eig(res.pca, addlabels=TRUE, ylim=c(0,50))

var <- get_pca_var(res.pca)
var

head(var$coord)
head(var$cos2)
head(var$contrib)
head(var$cor)

## Plot variables - Correlation Circle
win.graph(width=7.0, height=5.5)
fviz_pca_var(res.pca, col.var="black")

## Quality of representation
head(var$cos2, 4)

library("corrplot")
win.graph(width=7.0, height=5.5)
corrplot(var$cos2, is.corr=FALSE)

# Total cos2 for variables on PC1 and PC2
win.graph(width=7.0, height=5.5)
fviz_cos2(res.pca, choice="var", axes=1:2)

# Color by cos2 values: quality on factor map
win.graph(width=7.0, height=5.5)
fviz_pca_var(res.pca, col.var="cos2",
      gradient.cols = c("#00AFBB", "#E7B800", "#FC4E07"),
      repel = TRUE   # Avoid text overlap"
      )

# Contribution of variables to PCs
head(var$contrib, 4)

# Check calculations/definitions
sum(var$contrib[,1])
var$coord[1,1]^2
var$cos2[1,1]
var$coord[1,1]^2 / sum(var$coord[,1]^2)
var$cos2[1,1] / sum(var$cos2[,1])
var$contrib[1,1]

win.graph(width=7.0, height=5.5)
corrplot(var$contrib, is.corr=FALSE)

# Barchart of contributions by variable for PC1, PC2
#    includes highest 10 vars for each PC
win.graph(width=7.0, height=5.5)
fviz_contrib(res.pca, choice="var", axes=1, top=10)
win.graph(width=7.0, height=5.5)
fviz_contrib(res.pca, choice="var", axes=2, top=10)

## Combined Contribution to first 2 PCs
contrib.12 <- (eig.val[1] * var$contrib[,1] + eig.val[2] * var$contrib[,2]) /
                (eig.val[1] + eig.val[2])
contrib.12

# Color by contrib values: quality on factor map
win.graph(width=7.0, height=5.5)
fviz_pca_var(res.pca, col.var="contrib",
      gradient.cols = c("#00AFBB", "#E7B800", "#FC4E07")
      )


# Color by custom continuous variable
set.seed(123)
my.cont.var <- rnorm(4)
win.graph(width=7.0, height=5.5)
fviz_pca_var(res.pca, col.var=my.cont.var,
      gradient.cols = c("blue", "yellow", "red"),
      legend.title = "Cont. Var."
      )


# Color by groups
# Create 3 groups of vars (centers=3)
set.seed(123)
res.km <- kmeans(var$coord, centers=3, nstart=25)
grp <- as.factor(res.km$cluster)

win.graph(width=7.0, height=5.5)
fviz_pca_var(res.pca, col.var=grp,
      palette = c("#0073C2FF", "#EFC000FF", "#868686FF"),
      legend.title = "Cluster"
      )

# Dimension Description
res.desc <- dimdesc(res.pca, axes=c(1,2), proba=0.05)
Descriptions of dimensions 1 and 2
res.desc$Dim.1
res.desc$Dim.2

#### Analysis of Individuals

ind <- get_pca_ind(res.pca)
ind

head(ind$coord)
head(ind$cos2)
head(ind$contrib)

# Graph of Individuals
# Color by cos2 values 
win.graph(width=7.0, height=5.5)
fviz_pca_ind(res.pca, col.ind="cos2",
      gradient.cols = c("#00AFBB", "#E7B800", "#FC4E07"),
      repel = TRUE   # Avoid text overlap"
      )

# Change point size
win.graph(width=7.0, height=5.5)
fviz_pca_ind(res.pca, pointsize="cos2",
      pointshape=21, fill = "#E7B800",
      repel = TRUE   # Avoid text overlap"
      )

# Color and pointsize by cos2 values 
win.graph(width=7.0, height=5.5)
fviz_pca_ind(res.pca, col.ind="cos2", pointsize="cos2",
      gradient.cols = c("#00AFBB", "#E7B800", "#FC4E07"),
      repel = TRUE   # Avoid text overlap"
      )

# Barchart of quality of representation
win.graph(width=7.0, height=5.5)
fviz_cos2(res.pca, choice="ind")

# Total contribution on PC1 and PC2
win.graph(width=7.0, height=5.5)
fviz_contrib(res.pca, choice="ind", axes=1:2)

# Color by custom continuous variable 
set.seed(123)
my.cont.var <- rnorm(23)
win.graph(width=7.0, height=5.5)
fviz_pca_ind(res.pca, col.ind=my.cont.var,
      gradient.cols = c("blue", "yellow", "red"),
      legend.title = "Cont. Var."
      )