#install.packages('readxl')
#install.packages('stringr')

library(readxl)
library(stringr)

entropy = function(p)
{
  sum(ifelse(p == 0, 0, -p*log(p,2)))
}

data = read_excel('AkinatorTabela.xlsx')


#name_encoding = model.matrix(~Name-1, data = data)
#colnames(name_encoding) = sapply(colnames(name_encoding), function(x) paste('Is the person ', str_sub(x,5), '?', sep = ''))
#data = cbind(data[,1:(ncol(data) - 1),], name_encoding, data[,ncol(data)])

par(mar=c(5,13,2,2))

sub_data = data
answer = 0
n_questions = ncol(data) - 2
while(answer != -1)
{
  print(paste('Current entropy = ',entropy(sub_data$Prob)))
  if(entropy(sub_data$Prob) == 0)
  {
    print(paste('The person is', sub_data$Name[which.max(sub_data$Prob)]))
    answer = -1
  }
  else
  {
      barplot(sub_data$Prob, horiz = T, names.arg = sub_data$Name, col = 'salmon', main = 'Most likely person',xlab = 'Probability', las = 2)
      question_information = apply(sub_data[,2:(n_questions + 1)], 2, function(x) entropy(c(sum(data$Prob[x==0]), sum(data$Prob[x==1]))))
      #print(sort(question_information, decreasing = TRUE))
      readline(prompt="Press [enter] to continue")
      barplot(sort(question_information), horiz = T, col = 'skyblue', main = 'Questions:', xlab = 'Average information gain', las = 2)
      max_information = max(question_information)
      best_questions = names(question_information)[question_information == max_information]
      question = sample(best_questions, 1)
      answer = as.numeric(readline(prompt = question))
      sub_data = sub_data[sub_data[,question] == answer,]
      sub_data$Prob = sub_data$Prob / sum(sub_data$Prob)
    }
}