In benjaminguinaudeau/fifaindex: What the Package Does (One Line, Title Case)

knitr::opts_chunk$set(echo = TRUE)

pacman::p_load(tidyverse, furrr, fifaindex)

plan(multiprocess, workers = 2)

source("R/league_helpers.R")

Get_all_leagues

Get all league_id

league_ids <- get_all_league_ids()

save(league_ids, file = "data/league/potential_league_id.Rdata")
league_ids <- league_ids %>%
  filter(exists) %>%
  pull(potential_ids)

save(league_ids, file = "data/league/league_ids.Rdata")

Get each league

load("data/league/league_ids.Rdata")

league_data <-  league_ids %>%
  imap(~{print(.y);get_league(.x)})

league_data <- league_data %>%
  reduce(bind_rows) %>%
  select(-x, -team_rating) %>%
  mutate(team_link = paste0("https://www.fifaindex.com", team_link)) %>%
  mutate(league_id = url %>% str_extract("(?<=league\\=)\\d+") %>% as.numeric) %>%
  mutate(year = str_extract(period_date, "\\d+")) %>%
  glimpse %>%
  select(league_id, league, year, name, team_id, att, mid, def, ovr) %>%
  left_join(fifa_monks_dict, by = c("league_id" = "fifa_id")) %>%
  filter(!is.na(monks_id)) %>%
  glimpse

save(league_data, file = "data/league/league_data.Rdata")

Merge with monks

load("data/league/league_data.Rdata")
load("data/league/league_teams.Rdata")

Transform data from monks

# Number of teams
league_data %>%
  count(name, sort = T)

monks_data <- league_teams %>%
  select(-national_team, -logo_path) %>% 
  mutate(longest =  longest(team_name)) %>%
  filter(league_id %in% unique(league_data$monks_id)) %>%
  select(league_id, team_id, team_name, short_code, longest, 
         monks_country_id = country_id, founded, monks_venue_id = venue_id, twitter) %>%
  distinct %>%
  glimpse

Fix names issues in fifa data

prep_league_data <- league_data %>%
  filter(year > 2014) %>%
  distinct(league_id, league, monks_id, name, team_id) %>%
  mutate(name = fix_team_names(name, league_id)) %>%
  filter(name != "") %>%
  glimpse()

Join fifa and monks

joined_data <- prep_league_data %>%
  left_join(monks_data, c("monks_id" = "league_id", 
                          "name" = "team_name")) %>%
  split(is.na(.$team_id.y)) %>%
  imap_dfr(~{
    if(.y == "FALSE") return(.x)

    .x %>%
      select(league_id, league, monks_id, name, team_id = team_id.x) %>%
      mutate(longest =  longest(name)) %>%
      left_join(monks_data, c("monks_id" = "league_id", 
                              "longest" = "longest")) %>%
      select(-team_name)
  }) %>%
  select(fifa_league_id = league_id, monks_league_id = monks_id,
         league,
         fifa_team_id = team_id.x, monks_team_id = team_id.y, 
         -longest,
         everything()
         ) 

# Get those that were not joined
not_joined <- joined_data %>% #1136
  filter(is.na(monks_team_id))

# Find the monks league data given a fifa league_id
joined_data %>%
  filter(league_id %in% c(308)) %>%
  count(league_id, monks_id) %>% 
  filter(league_id %in% monks_league$monks_id) %>%
  left_join(monks_league, by = c("league_id" = "monks_id")) %>%
  arrange(league_id.y) %>%
  select(league_id, league_id.y, everything())

fifa_monks_team_dic <- joined_data %>%
  select(fifa_team_id, monks_team_id, fifa_league_id, monks_league_id) %>%
  unique

save(fifa_monks_team_dic, file = "data/league/fifa_monks_team_dic.Rdata")

fifa_league_data <- league_data %>%
  rename(fifa_team_id = team_id, fifa_league_id = league_id) %>%
  left_join(fifa_monks_team_dic, by = c("fifa_team_id", "fifa_league_id")) %>%
  #rename()
  left_join(league_teams, by = c("monks_team_id" = "team_id", 
                                 "monks_league_id" = "league_id")) %>%
  unique %>%
  select(fifa_league_id, monks_league_id,
         league, year,
         fifa_team_id, monks_team_id, 
         team_name, fifa_team_name = name,
         ovr, att, mid, def,
         country_id, founded, venue_id = venue_id,
         everything(),
         -national_team, -logo_path, -legacy_id, -current_season_id, -monks_id 
         ) 

fifa_league_data %>%
  glimpse
save(fifa_league_data, file = "data/league/fifa_league_data.Rdata")