R/CongestionDetection.R

In atousaz/Dcongestion: Spatiotemporal congestion detection

#' @title Data driven congestion detection.
#' @name Data Driven congestion detection.
#' @param dataset including speed signal, change point probabilities
#' @return Spatiotemporal congestions
#' @author Atousa zarindast
#' @import dplyr
#' @return classified dataframe
#' @examples
#' \dontrun{
#' library("Dcongestion")
#' dataset <- load(package = "Dcongestion", "dataset")
#' Congestion_detection(dataset)}
library(dplyr)
Congestion_detection <- function(dataset) {


  dataset <- as.data.frame(dataset[,-1])
  dataset$cp <- 0
  dataset$group <- "NA"

  n = 1

  while (n < (nrow(dataset) - 1)) {
    n = n + 1
    #print(n)
    if ((n + 1) == (nrow(dataset) + 1)) {
      break()
    }

    else
      if (!dataset$Prob[n]  <  0.01) {
        for (j in (n  +  1):nrow(dataset)) {
          #print(n)


          if (dataset$Prob[j]  <  0.01) {
            d  =  max(dataset$Prob[n:j  -  1])
            #print(d)
            t  <-  dataset$Prob[n:j  -  1]  ==  d
            #print(t[-1])
            dataset$cp[n:j  -  1][which(t  ==  TRUE)]  <-  1


            break()

            #print(n)

          }

        }

        n <- j
      }

  }





  rm(d)
  rm(j)
  rm(n)
  rm(t)



  counter = 1
  l = 1
  for (i in 1:nrow(dataset)) {
    if (dataset$cp[i] == 1) {
      dataset$group[l:i] <- counter
      #print(counter)
      counter = counter + 1
      l = i + 1
      #print(l)
    }
    # check<-dataset$cp[i+1:nrow(dataset)]==0
    # if (dataset$cp[i]==1 & unique(check)=="TRUE"){
    #
    #   dataset$group[i:nrow(dataset)]<-counter+1
    # }
    #
  }


  dataset$group[dataset$group == "NA"] <- counter
  rm(counter, i, l)
  #write.csv(dataset,"1485655352/result/grouping_test.csv")
  total = NULL
  group = unique(dataset$group)
  for (g in group) {
    fil_data <-
      dataset %>% filter(group == g) %>% mutate(
        Q15 = quantile(Speed, 0.15),
        Q50 = quantile(Speed, 0.50),
        Q85 = quantile(Speed, 0.85)
      )
    total <- rbind(total, fil_data)

  }

  rm(g, group, fil_data)
  #write.csv(total,"total.csv")

  clusters <- kmeans(total[, c('Q15', 'Q50', 'Q85')], 2)

  d <- cbind(cluster = clusters$cluster, total)

  m <-
    d %>% group_by(cluster) %>% mutate(avg = mean(Speed)) %>% group_by(cluster, avg) %>%
    summarize()
  m <- as.data.frame(m)
  clu <-
    m %>% filter(avg == min(avg)) %>% select(cluster)#finding the cluster

  d$visual <- ifelse(d$cluster == clu$cluster, d$Speed, 0)
  rm(clusters, m)

  dataset <- d
  rm(d, total)



  return(dataset)
}