mtt_haum/code/12_dfgs-case-clusters.R

# 12_dfgs-case-clusters.R
#
# content: (1) Read data
#          (2) Export DFGs for clusters
#
# input:  results/user-navigation.RData
# output: ../thesis/figures/dfg_cases_cluster1_R.pdf
#         ../thesis/figures/dfg_cases_cluster2_R.pdf
#         ../thesis/figures/dfg_cases_cluster3_R.pdf
#         ../thesis/figures/dfg_cases_cluster4_R.pdf
#         ../thesis/figures/dfg_cases_cluster5_R.pdf
#         ../thesis/results/dfgs_case-cluster.RData
#
# last mod: 2024-04-17

# setwd("C:/Users/nwickelmaier/Nextcloud/Documents/MDS/2023ss/60100_master_thesis/analysis/")

#--------------- (1) Read data ---------------

load("results/user-navigation.RData")

dat <- res

dat$start <- as.POSIXct(dat$date.start)
dat$complete <- as.POSIXct(dat$date.stop)

alog <- bupaR::activitylog(dat[dat$cluster == 4, ],
                            case_id     = "case",
                            activity_id = "item",
                            resource_id = "path",
                            timestamps  = c("start", "complete"))

processmapR::trace_explorer(alog, n_traces = 25)

tr <- bupaR::traces(alog)
tab <- table(tr$absolute_frequency)

tab[1] / nrow(tr)

alog |> edeaR::filter_infrequent_flows(min_n = 5) |> processmapR::process_map()

#--------------- (2) Export DFGs for clusters ---------------

mycols <- c("#3CB4DC", "#FF6900", "#78004B", "#91C86E", "#434F4F")
cl_names <- c("Scanning", "Exploring", "Flitting", "Searching", "Info")

ns <- c(30, 20, 10, 5, 30)
#ns <- c(20, 20, 20, 5, 20)

for (i in 1:5) {

  alog <- bupaR::activitylog(dat[dat$cluster == i, ],
                             case_id     = "case",
                             activity_id = "item",
                             resource_id = "path",
                             timestamps  = c("start", "complete"))
  if (i %in% c(4, 5)) {

    dfg <- processmapR::process_map(edeaR::filter_infrequent_flows(alog, min_n = ns[i]),
      type_nodes = processmapR::frequency("relative", color_scale = "Greys"),
      sec_nodes  = processmapR::frequency("absolute"),
      type_edges = processmapR::frequency("relative", color_edges = mycols[i]),
      sec_edges  = processmapR::frequency("absolute"),
      render     = FALSE)

  } else {

    dfg <- processmapR::process_map(edeaR::filter_infrequent_flows(alog, min_n = ns[i]),
      type_nodes = processmapR::frequency("relative", color_scale = "Greys"),
      sec_nodes  = processmapR::frequency("absolute"),
      type_edges = processmapR::frequency("relative", color_edges = mycols[i]),
      sec_edges  = processmapR::frequency("absolute"),
      rankdir    = "TB",
      render     = FALSE)

  }

  processmapR::export_map(dfg,
    file_name = paste0("../thesis/figures/dfg_cases_cluster_", cl_names[i], "_R.pdf"),
    file_type = "pdf")
}

## Black and white
for (i in 1:5) {

  alog <- bupaR::activitylog(dat[dat$cluster == i, ],
                             case_id     = "case",
                             activity_id = "item",
                             resource_id = "path",
                             timestamps  = c("start", "complete"))
  if (i %in% c(4, 5)) {

    dfg <- processmapR::process_map(edeaR::filter_infrequent_flows(alog, min_n = ns[i]),
      type_nodes = processmapR::frequency("relative", color_scale = "Greys"),
      sec_nodes  = processmapR::frequency("absolute"),
      type_edges = processmapR::frequency("relative", color_edges = "black"),
      sec_edges  = processmapR::frequency("absolute"),
      render     = FALSE)

  } else {

    dfg <- processmapR::process_map(edeaR::filter_infrequent_flows(alog, min_n = ns[i]),
      type_nodes = processmapR::frequency("relative", color_scale = "Greys"),
      sec_nodes  = processmapR::frequency("absolute"),
      type_edges = processmapR::frequency("relative", color_edges = "black"),
      sec_edges  = processmapR::frequency("absolute"),
      rankdir    = "TB",
      render     = FALSE)

  }

  processmapR::export_map(dfg,
    file_name = paste0("../thesis/figures/dfg_cases_cluster_", cl_names[i], "_R_bw.pdf"),
    file_type = "pdf")
}

# What data is used and how many traces are unique

tr_unique   <- numeric(5)
perc_filter <- numeric(5)
n_cases     <- numeric(5)

for (i in 1:5) {

  alog <- bupaR::activitylog(dat[dat$cluster == i, ],
                             case_id     = "case",
                             activity_id = "item",
                             resource_id = "path",
                             timestamps  = c("start", "complete"))

  cs <- bupaR::cases(alog)

  cs_filtered <- edeaR::filter_infrequent_flows(alog, min_n = ns[i]) |>
    bupaR::cases()

  n_cases[i] <- nrow(cs_filtered)
  perc_filter[i] <- n_cases[i] / nrow(cs)

  tr <- bupaR::traces(alog)

  sum_tr <- sum(tr$absolute_frequency == 1)
  tr_unique[i] <- sum_tr / nrow(tr)

  #infreq_tr <- names(which(table(cs$trace) == 1))
  #infreq_cs <- cs$case[cs$trace %in% infreq_tr]
  #cs_filtered[i] <- length(infreq_cs) / length(cs$case)

}

save(ns, n_cases, tr_unique, perc_filter,
     file = "../thesis/results/dfgs_case-cluster.RData")