# importing libraries
library (factoextra)
library (cluster)
library("ggpubr")

# importing data
data <- read.csv("C:\\Users\\ramta\\Desktop\\Kvartiry_Ufa.csv")

# viewing head of data
head(data)

# droping X column, cuz it is repeating index, which needless
data = subset(data, select = -c(X))

#checking head of data again
head(data)

# removing rows with missing values to performance clustering
data <- na.omit(data)

# plotting data to see the spread and correlation
plot(data)

# data is too big, plotting specific columns
plot(data$total_area, data$last_price)
points(data$citycenter_near, data$last_price, col="red")
points(data$parks_around3, data$last_price, col="green")
points(data$living_area, data$last_price, col="blue")

# take a closer look to column citycenter_near
plot(data$citycenter_near, data$last_price, col="red")

# scale each variable to have a mean of 0 and sd of 1
data <- scale(data)

# view head of scaled data
head(data)

#make this example reproducible
set.seed(1)

# figuring out optimal numbers of clusters
fviz_nbclust(data, kmeans, nstart=10, method = "wss")

# distortion decreasing at point 8
# perform k-means clustering with k = 8 clusters
km <- kmeans(data, centers = 4, nstart = 25)

# view results
print(km)

#plot results of final k-means model
fviz_cluster(km, data = data)

km$centers

plot(data, col=km$cluster)
points(km$centers, col=1:4, pch=8)