Marvel.qmd

# Marvel Universe Network

```{r}
#| label: load_libraries
#| echo: false
#| message: false
#| warning: false

library(rvest)
library(tidyverse)
library(tidygraph)
library(visNetwork)
library(janitor)
```

## Récupération des données depuis Wikipedia

```{r}
#| label: scrap

url1 <- "https://en.wikipedia.org/w/index.php?title=Marvel_Cinematic_Universe:_Phase_One&oldid=1096778874"
url2 <- "https://en.wikipedia.org/w/index.php?title=Marvel_Cinematic_Universe:_Phase_Two&oldid=1097878759"
url3 <- "https://en.wikipedia.org/w/index.php?title=Marvel_Cinematic_Universe:_Phase_Three&oldid=1096776335"

urls <- c(url1, url2, url3)

get_table <- function(wiki_url) {
  
  tib <- wiki_url |> 
    read_html() |> 
    html_elements(".wikitable") |> 
    html_table(na.strings = "") |> 
    pluck(3) |> 
    row_to_names(row_number = 1) |> 
    pivot_longer(
      cols = -c("Character"), 
      names_to = "Movie", 
      values_to = "Actor", 
      values_drop_na = TRUE) |> 
    mutate(
      Actor = str_remove_all(Actor, "\\[|\\]"),
      Actor = str_remove_all(Actor, "[:digit:]"),
      Actor = str_remove(Actor, "[C|V]$"),
      Actor = str_replace_all(Actor, "([a-z])([A-Z])", "\\1 \\2"),
      Character = str_remove_all(Character, '\\\"'),
      Character = str_replace_all(Character, "([a-z])([A-Z])", "\\1 \\2"),
      Character = str_remove(Character, "OS$"),
      Character = str_trim(Character),
      Movie = str_replace_all(Movie, "([a-z]|:)([A-Z])", "\\1 \\2"),
      Movie = str_trim(Movie),
      Movie = recode(Movie, "Thor" = "Thor 1", "Iron Man" = "Iron Man 1")
    ) |> 
    select(-Actor)
  
  return(tib)

}

df_marvel <- map(urls, get_table) |> 
  list_rbind() # map_dfr(urls, get_table) is deprecated
```

## Analyse de données exploratoire

```{r}
#| label: analysis

# Personnage apparaissant le plus souvent (en terme de films)

df_marvel |>
  count(Character, sort = TRUE)

# Films faisant intervenir le plus de personnages

df_marvel |>
  count(Movie, sort = TRUE)

```

## Création d'un graphe

```{r}
#|label: graph
network <- df_marvel |> 
  as_tbl_graph()

class(network)

# Mesure de centralité, par exemple le degré associé à un noeud

network |> 
  activate(nodes) |> # on considère la table des noeuds ou "vertices"
  mutate(degree = centrality_degree()) |> 
  as_tibble() |>
  arrange(desc(degree))

```

## Interactivité avec le paquet `visNetwork`

Pour aller plus loin dans les options ne pas hésiter à visiter [visNetwork, an R package for interactive network visualization](https://datastorm-open.github.io/visNetwork/) 

```{r}
#| label: visNetwork

vis_network <- network |> 
  mutate(group = if_else(
    condition = name %in% unique(df_marvel$Character),
    true = "Character",
    false = "Movie"
  )) |> 
  toVisNetworkData()

visNetwork(
  nodes = vis_network$nodes,
  edges = vis_network$edges,
  width = "100%",
  height = "600px",
  main = "L'univers filmographique Marvel") |> 
  visLayout(randomSeed = 1000) |> 
  addFontAwesome() |> 
  visGroups(
    groupname = "Movie",
    shape = "icon",
    icon = list(code = "f008", color = "darkblue")
  ) |> 
  visGroups(
    groupname = "Character",
    shape = "icon",
    icon = list(code = "f007", color = "red")
  ) |> 
  visOptions(
    highlightNearest = list(enabled = TRUE, hover = TRUE),
    nodesIdSelection = TRUE
  ) |> 
  visInteraction(navigationButtons = TRUE)

```