R/prepare_network_edges.R

In ryanjessop/ClickViz: Clickstream Visualisations

#' Prepare the network edges
#'
#' Transform and clean the clickstream data into transition probabilites, p_{i,j}.
#' Output dataframe of network edges.
#'
#' @param data Clickstream data
#' @param alpha Display only p_{i,j} > alpha
#' @return Network edges data
#' @examples
#' cls <- readClickstreams(file)
#' prepare_network_edges(cls)
prepare_network_edges <- function(data,
                                  alpha = 0.01) {
  mc <- clickstream::fitMarkovChain(data)
  transition_matrix <- as.data.frame(t(mc@transitions$`1`))
  state <- rownames(transition_matrix)
  add_state_column <- cbind(state, transition_matrix)

  tidy_transitions <- tidyr::gather(add_state_column,
                                    as.vector(state),
                                    key = 'next_state',
                                    value = 'transition_p') %>%
    dplyr::filter(transition_p > alpha) %>%
    dplyr::mutate(join_col = round(transition_p, 2))

  max_p <- max(tidy_transitions$join_col)
  count_sequence <- seq(0, max_p, 0.01)

  color_map <- tibble::tibble(
    greyscale = grDevices::gray.colors(
      max_p * 100 + 1,
      start = 0,
      end = 0,
      alpha = count_sequence
    ),
    join_col = count_sequence
  )

  network_edges <-
    dplyr::inner_join(tidy_transitions, color_map, by = "join_col")

  return(network_edges)
}