# This code is used to group objects using the method k-means
# Install the following packages:
install.packages("ggsignif", type="win.binary") 
install.packages("ggpubr", type="win.binary") 
install.packages("factoextra")
install.packages("dplyr")
install.packages("writexl")
# Import data file (File -> Import dataset -> From Excel...)
# The contents of the first column should be row name

df<-data.frame(Data_UE_2_,row.names = TRUE)
mydata <- scale(df) # variables normalisation

# Load factoextra package
library(factoextra)
# Use the fviz_nbclust functions to determine the optimal number of clusters
fviz_nbclust(mydata, kmeans, method = "silhouette") 
#Group similar objects with k-means methods for the indicated number of clusters
km.res <- kmeans(mydata, 4, nstart = 25)         
print(km.res) # Show cluster results
aggregate(df, by=list(cluster=km.res$cluster), mean) #average values of variables in clusters

library(RColorBrewer)
display.brewer.all()
# Clusters visualisation
fviz_cluster(km.res, data = mydata,
             palette = c("Accent"), 
             ellipse.type = "euclid", 
             star.plot = TRUE, 
             repel = TRUE, #Non-overlapping labels
             ggtheme=theme_minimal())

#adding the cluster number to the  dataset
clusters <- cbind(df, cluster = km.res$cluster) 

# saving data to excel
library("dplyr")
library("writexl")


#row name as column
data<-tibble::rownames_to_column(clusters, "Country")
write_xlsx(data, path = "clusters.xlsx") 

#The end

## Agglomeration clustering - Ward's method (the optimal number of groups should be determined beforehand)
hc.res <- eclust(mydata, "hclust", k = 3, hc_metric = "euclidean", 
                 hc_method = "ward.D2")
# Creating a dendrogram
fviz_dend(hc.res, show_labels = TRUE,
          palette = "jco", as.ggplot = TRUE)

# adding the cluster number to the output dataset
clusters22 <- cbind(df, cluster = hc.res$cluster)