# Displaying High-Dimensional Data in R
#
# Get the Iris dataset
#
Data <- iris
# 3x50 irises, 4 variables plus iris species
str(Data)
summary(Data)
#
Data <- na.omit(Data)
#
Gcol <- as.numeric(Data[ ,5])
#
# Remove "species" from the data
#
IRIS <- Data[ ,-5]
#
## install.packages("FactoMineR")
## install.packages("factoextra")
#
library(FactoMineR)
library(factoextra)
#
IRIS_PCA <- prcomp(IRIS)
#
# Plot variables in the PCA space
#
plot(IRIS_PCA$x[ ,1:2], col=Gcol, pch=16)
#
#
# PCA plot of iris data colored by species with ellipses added
#
iris.pca <- PCA(iris[,-5], graph = FALSE)
fviz_pca_ind(iris.pca,
             geom.ind = "point", # show points only (nbut not "text")
             col.ind = iris$Species, # color by groups
             palette = c("#00AFBB", "#E7B800", "#FC4E07"),
             addEllipses = TRUE, # Concentration ellipses
             legend.title = "Groups"
             )
#
# Sammon's nonlinear mapping
#
## install.packages("MASS")
## install.packages("magrittr")
## install.packages("dplyr")
## install.packages("ggpubr")
library(MASS) 
library(magrittr) 
library(dplyr) 
library(ggpubr) 
#
IRIS2 <- IRIS
IRIS2[143,2] <- IRIS2[143,2] + 0.05
#
Dist_IRIS2 <- dist(IRIS2)
mds_SAM <- sammon(Dist_IRIS2, niter=1000)
plot(mds_SAM$points, pch=16, col=Gcol, main="Sammon map")
#
# Kruskal nonlinear mapping
#
mds_KR <- isoMDS(Dist_IRIS2, maxit=1000)
plot(mds_KR$points, pch=16, col=Gcol, main="Kruskal map")
#
# t-SNE
#
## install.packages("tsne")
library(tsne)
IRIS_tsne <- tsne(IRIS, initial_dims = 2)
IRIS_tsne <- data.frame(IRIS_tsne)
attributes(IRIS_tsne)
#
plot(IRIS_tsne$X1, IRIS_tsne$X2, col=Gcol, pch=16)
#
# Another t-SNE test
#
IRIS_tsne <- tsne(IRIS, initial_dims=3, max_iter = 2000, perplexity=30)
IRIS_tsne <- data.frame(IRIS_tsne)
plot(IRIS_tsne$X1, IRIS_tsne$X2, col=Gcol, pch=16)
#
# UMAP
#
## install.packages("umap")
library(umap)
#
IRIS.umap <- umap(IRIS, n_neighbors = 5)
plot(IRIS.umap$layout, col=Gcol, pch=16)
#
x <- runif(50)
IRIS.umap <- umap(IRIS, n_neighbors = 10)
plot(IRIS.umap$layout, col=Gcol, pch=16)
#
x <- runif(50)
IRIS.umap <- umap(IRIS, n_neighbors = 10)
plot(IRIS.umap$layout, col=Gcol, pch=16)
#
x <- runif(50)
IRIS.umap <- umap(IRIS, n_neighbors = 15)
plot(IRIS.umap$layout, col=Gcol, pch=16)
#
# Parallel Coordinates
#
parcoord(IRIS, col = Gcol, var.label = TRUE, main = "iris dataset")
#