#install.packages("clusterCrit")
library(clusterCrit)
library(cluster)

# HEADERS
iris_headers = c("sepal_length", "sepal_width", "petal_length", "petal_width", "class")
wine_headers = c("class", "Alcohol", "Malic_acid", "Ash", "Alcalnity_of_ash", "Magnesium", "Total_phenols", "Flavanoids", "Nonflavanoid_phenols", "Proanthocyanins", "Color_intensity", "Hue", "OD280/OD315_of_diluted_wines", "Proline")
glass_headers = c("Id", "Refractive_index", "Na", "Mg", "Al", "Si", "K", "Ca", "Ba", "Fe", "class")
diabetes_headers = c("Pregnant_times", "Plasma_glucose", "Blood_pressure", "Triceps_skin_thickness", "Serum_insulin", "BMI",
                     "Diabetes_pedigree", "Age", "class")

get_data <- function(datasetName, dataset) {
  switch(datasetName,
         "iris" = {
           data <- dataset[ ,1:4]
         },
         "wine" = {
           data <- dataset[ ,2:13]
         },
         "glass" = {
           data <- dataset[ ,2:10]
         },
         "diabetes" = {
           data <- dataset[ ,1:8]
         })
  return (data)
}

get_class <- function(datasetName, dataset) {
  switch(datasetName,
         "iris" = {
           class <- dataset[, 5]
         },
         "wine" = {
           class <- dataset[, 1]
         },
         "glass" = {
           class <- dataset[, 11]
         },
         "diabetes" = {
           class <- dataset[, 9]
         })
  return (class)
}

load_dataset <- function(filename, headers) {
  dataset <- read.csv(filename, header = FALSE)
  colnames(dataset) <- headers
  return (dataset)
}

calculateCriterias <- function(data, cl) {
  intCrit <- intCriteria(data, cl$cluster, c("Davies_Bouldin", "Dunn", "Silhouette"))
  # extCrit <- extCriteria(classes, cl$cluster, c("Rand"))
  return(intCrit)
  # return(c(intCrit))
}

clasterise_kmeans <- function(dataset, numberOfClusters){
  return(kmeans(dataset, numberOfClusters))
}

clasterise_pam <- function(dataset, numberOfClusters){
  return(pam(dataset, numberOfClusters))
}

#DATASETS
# iris <- load_dataset("iris.csv", iris_headers)
# wine <- load_dataset("wine.csv", wine_headers)
# glass <- load_dataset("glass.csv", glass_headers)
diabetes <- load_dataset("diabetes.csv", diabetes_headers)

# iris_data = get_data("iris", iris)
# iris_class = get_class("iris", iris)
# wine_data = get_data("wine", wine)
# wine_class = get_class("wine", wine)
# glass_data = get_data("glass", glass)
# glass_class = get_class("glass", glass)
diabetes_data = get_data("diabetes", diabetes)
diabetes_class = get_class("diabetes", diabetes)

#MAIN
data = diabetes_data
class = diabetes_class
number_of_clusters = 3

data_matrix <- matrix(unlist(data), ncol = length(data), byrow = TRUE)
kmeans_cl <- clasterise_kmeans(data, number_of_clusters)
pam_cl <- clasterise_pam(data, number_of_clusters)
kmeans_criterias <-calculateCriterias(data_matrix, kmeans_cl)
pam_criterias <-calculateCriterias(data_matrix, pam_cl)
print(kmeans_criterias)
print(pam_criterias)