## ----include = FALSE----------------------------------------------------------
knitr::opts_chunk$set(collapse = TRUE, comment = "#>", fig.width = 6,
                      fig.height = 4)
library(simOutrank)

## ----base---------------------------------------------------------------------
traces <- as_traces(illustrative_log, "case_id", "activity", "timestamp")

criteria <- function(activity_sim = 0.8, activity_indiff = 0.7,
                     w_activity = 0.2) {
  list(
    crit_activity_profile(weight = w_activity, indifference = activity_indiff,
                          similarity = activity_sim, veto = 0.4),
    crit_edit_distance(weight = 0.2, similarity = 2, indifference = 3,
                       veto = 6),
    crit_nominal("status", weight = 0.3),
    crit_nominal("satisfaction", weight = 0.3)
  )
}

## ----rand---------------------------------------------------------------------
rand_index <- function(a, b) {
  agree <- 0L; n <- length(a)
  for (i in seq_len(n - 1L)) for (j in (i + 1L):n) {
    agree <- agree + ((a[i] == a[j]) == (b[i] == b[j]))
  }
  agree / choose(n, 2)
}

baseline <- cluster_traces(outrank_similarity(traces, criteria()),
                           k = 4, seed = 42)$memberships

## ----quantile-----------------------------------------------------------------
crit_fixed    <- crit_activity_profile(0.2, indifference = 0.7, similarity = 0.8)
crit_quantile <- crit_activity_profile(0.2, indifference = as_quantile(0.5),
                                       similarity = as_quantile(0.8))
crit_quantile

## ----threshold-sweep----------------------------------------------------------
grid <- seq(0.75, 0.95, by = 0.05)  # kept above the indifference of 0.70
sens <- t(vapply(grid, function(s) {
  sim <- outrank_similarity(traces, criteria(activity_sim = s))
  memb <- cluster_traces(sim, k = 4, seed = 42)$memberships
  c(mean_S = mean(sim$S[upper.tri(sim$S)]),
    rand   = rand_index(memb, baseline))
}, numeric(2)))
data.frame(similarity = grid, sens)

## ----weight-sweep-------------------------------------------------------------
weights <- c(0.05, 0.2, 0.5, 1)
w_sens <- vapply(weights, function(w) {
  sim <- suppressMessages(outrank_similarity(traces,
                                             criteria(w_activity = w)))
  rand_index(cluster_traces(sim, k = 4, seed = 42)$memberships, baseline)
}, numeric(1))
data.frame(w_activity = weights, rand = w_sens)

## ----eigengap-----------------------------------------------------------------
sim <- outrank_similarity(traces, criteria())
gaps <- eigengap(sim, k_max = 8)
gaps[which.max(gaps$gap), ]

## ----kvalidity----------------------------------------------------------------
if (requireNamespace("clValid", quietly = TRUE)) {
  t(vapply(2:6, function(k) {
    cl <- cluster_traces(sim, k = k, seed = 42)
    c(k = k, validate_clusters(cl))
  }, numeric(3)))
}