BelutListik.qmd

---
title: "Analysing Fantasy Premier League data in R: Belut Listrik"
author: "Arif P. Sulistiono / @arifpras"
format: html
editor: visual
---

# Prerequisites

This initial step covers clearing the environment, installing/loading related libraries that will be used through the analysis, and setting the working directory.

```{r pre-requisites}
# Remove all items in Environment!
# rm(list = ls())
# ls()

pacman::p_load(tidyverse,
               stats,
               ggsoccer,
               # utils,
               tidytext,
               rvest,
               corrr,
               jsonlite,
               # fuzzyjoin,
               stringr,
               viridis,
               tictoc,
               ggrepel)

options(tz = "Europe/London")

# Working directory
setwd("~/OneDrive - The University of Nottingham/BB_SideProject/BelutListrik")
```

# Building database

## FPL: 2020/2021 and 2021/2022

From the FPL's database, we want to get information on players' performances in each game week in two seasons. To obtain the information, we will perform 'web-scrapping' on following repositories: "<https://github.com/vaastav/Fantasy-Premier-League/tree/master/data/2020-21/gws>" and "<https://github.com/vaastav/Fantasy-Premier-League/tree/master/data/2021-22/gws>". *Note: download the RAW version.*

```{r db_fpl}
# Season 2020/2021

fpl20 <-
  "https://raw.githubusercontent.com/vaastav/Fantasy-Premier-League/master/data/2020-21/gws/merged_gw.csv"

(fpl01 <- read_csv(fpl20, show_col_types = FALSE) %>%
    mutate(season = "20/21"))

# Season 2021/2022

fpl21 <-
  "https://raw.githubusercontent.com/vaastav/Fantasy-Premier-League/master/data/2021-22/gws/merged_gw.csv"

(fpl02 <- read_csv(fpl21, show_col_types = FALSE) %>%
    mutate(season = "21/22"))

# Combine those seasons

(fpl03 <- fpl01 %>%
    bind_rows(fpl02))

# save(fpl01, fpl02, fpl03,
#      file = "01_db_fpl.rdata")
# 
# load("01_db_fpl.rdata")
```

## Aggregating the FPL's statistics

The objective is to understand the average points, minutes, ICT index, threats, creativity, and influence in the last two seasons.

```{r fpl_agg}
(
  fpl04 <- fpl03 %>%
    group_by(name) %>%
    summarise(
      avg_points = mean(total_points),
      avg_minutes = mean(minutes),
      avg_ict = mean(ict_index),
      avg_threat = mean(threat),
      avg_creativity = mean(creativity),
      avg_influence = mean(influence),
      .groups = "keep"
    ) %>%
    ungroup() %>%
    arrange(desc(avg_points))
)

fpl04 %>% select(-name) %>%
  tibble() %>%
  correlate() %>%
  network_plot(min_cor = 0.9)

# save(fpl04,
#      file = "02_db_agg.rdata")
# load("02_db_agg.rdata")
```

## Understat: 2020/2021 and 2021/2022

From the Understat's database, we want to get other information (e.g., xG, xA, shots, key passes, npg, npxG, xGChain, xGBuildUp). The 'web-scrapping' will be performed on following repositories: "<https://github.com/vaastav/Fantasy-Premier-League/tree/master/data/2020-21>" and "<https://github.com/vaastav/Fantasy-Premier-League/tree/master/data/2021-22/understat>". *Note: download the RAW version.*

```{r db_understat}
# Season 2020/2021

uds20 <-
  "https://raw.githubusercontent.com/arifpras/Fantasy-Premier-League/master/data/2020-21/understat/understat_player.csv"

(
  uds01 <- read_csv(uds20, show_col_types = FALSE) %>%
    mutate(
      season = "20/21",
      xG90 = xG / time * 90, # Expected goals per 90 minutes played
      xA90 = xA / time * 90,
      npxG90 = npxG / time * 90,
      xGChain90 = xGChain / time * 90,
      xGBuildup90 = xGBuildup / time * 90
    )
)

# Season 2021/2022

uds21 <-
  "https://raw.githubusercontent.com/arifpras/Fantasy-Premier-League/master/data/2021-22/understat/understat_player.csv"


(
  uds02 <- read_csv(uds21, show_col_types = FALSE) %>%
    mutate(
      season = "21/22",
      xG90 = xG / time * 90, # Expected goals per 90 minutes played
      xA90 = xA / time * 90,
      npxG90 = npxG / time * 90,
      xGChain90 = xGChain / time * 90,
      xGBuildup90 = xGBuildup / time * 90
    )
)

# Combine those seasons

(uds03 <- uds01 %>%
    bind_rows(uds02))

# save(uds01, uds02, uds03,
#      file = "03_db_understat.rdata")
# load("03_db_understat.rdata")
```

## Aggregating Understat's statistics

The objective is to understand the average xG (expected goals), xA (expected assist), npxG (non-penalty expected goals), xGChain (number of Expected Goals in which a player has participated), and xGBuildup (which players are part of passing chains that end in a shot but without taking into account the last two links -- shot and last pass -, thus measuring midfielders, centre-backs or full-backs who have an important influence on their team's possessions) in the last two seasons.

```{r understat_agg}
(
  uds04 <- uds03 %>%
    group_by(player_name, id) %>%
    summarise(
      avg_xG = mean(xG),
      #expected goals
      avg_xA = mean(xA),
      #expected assist
      avg_npxG = mean(npxG),
      #non-penalty expected goals
      avg_xGChain = mean(xGChain),
      #number of Expected Goals in which a player has participated
      avg_xGBuildup = mean(xGBuildup),
      #which players are part of passing chains that end in a shot but without taking into account the last two links – shot and last pass -, thus measuring midfielders, centre-backs or full-backs who have an important influence on their team's possessions,
      avg_xG90 = mean(xG90),
      #every 90 minutes
      avg_xA90 = mean(xA90),
      #every 90 minutes
      avg_npxG90 = mean(npxG90),
      #every 90 minutes
      avg_xGChain90 = mean(xGChain90),
      #every 90 minutes
      avg_xGBuildup90 = mean(xGBuildup90),
      #every 90 minutes
      .groups = "keep"
    ) %>%
    ungroup() %>%
    arrange(desc(avg_xGChain))
)

uds04 %>%
  select(avg_xG, avg_xA, avg_npxG, avg_xGChain, avg_xGBuildup) %>%
  tibble() %>%
  correlate() %>%
  network_plot(min_cor = 0.7)

# save(uds04,
#      file = "04_understat_agg.rdata")
# load("04_understat_agg.rdata")
```

## Merging the databases

To do so, we need to have another file. Another web-scrapping needs to be done on following repository <https://github.com/vaastav/Fantasy-Premier-League/blob/master/data/2021-22/id_dict.csv>. But, before that we have to make sure no duplicated players due to the changing in their positions.

```{r dictionary}
dict01 <-
  "https://raw.githubusercontent.com/arifpras/Fantasy-Premier-League/master/data/2021-22/id_dict.csv"

(dict02 <- readr::read_csv(dict01, show_col_types = FALSE))

(dict03 <- fpl01 %>% select(name, position) %>% distinct(name, position))
(dict04 <- fpl02 %>% select(name, position) %>% distinct(name, position))

(dict05 <- dict03 %>%
    bind_rows(dict04) %>%
    arrange(name) %>%
    distinct(name, position) %>% 
    right_join(
      dict02, by = c("name" = "FPL_Name")
    ))

# Check duplicated players based on their positions
dict05 %>% group_by(name) %>% filter(n()>1) %>% ungroup()

# Stuart Dallas: DEF
# Matt Ritchie: DEF
# Allan Saint-Maximin: MID
# Ainsley Maitland-Niles: MID

(dict06 <- dict05 %>%
    filter(!( # "!" means exclude
        (name == "Stuart Dallas" & position != "MID") |
        (name == "Matt Ritchie" & position != "MID") |
        (name == "Allan Saint-Maximin" & position != "FWD") |
        (name == "Ainsley Maitland-Niles" & position != "DEF") |
        (
          name %in% c(
            "Álvaro Fernández",
            "Cheikhou Kouyaté",
            "Fernando Luiz Rosa",
            "Nathan Broadhead",
            "Pierre-Emerick Aubameyang"
          )
        )
    )))

# Check duplicated players
dict06 %>% group_by(name) %>% filter(n()>1)

# Not in the EPL anymore
dict02 %>% anti_join(dict06, by = c("Understat_Name"))

# save(dict06,
#      file = "06_dictionary.rdata")
# load("06_dictionary.rdata")
```

Now, join all the databases.

```{r db01}
# FPL data and the dictionary

(
  fpl05 <- fpl04 %>%
    select(name, avg_points, avg_minutes, avg_ict, avg_influence) %>%
    right_join(dict02, by = c("name" = "FPL_Name")) %>% 
    select(-Understat_ID, -name, -Understat_Name)
)

# Understat data and the dictionary

(
  uds05 <- uds04 %>% 
    select(player_name, id, avg_xG, avg_xA, avg_xGChain) %>% 
    right_join(dict02, by = c("id" = "Understat_ID")) %>%
    select(-player_name) %>% 
    rename(
      Understat_ID = id
    ) %>% 
    relocate(
      Understat_Name, Understat_ID, FPL_Name, FPL_ID
    ) 
)

# Joining fpl05 and uds05

(
  db01 <- fpl05 %>% 
    left_join(uds05, by = c("FPL_ID")) %>% 
    relocate(
      Understat_Name, Understat_ID, FPL_Name, FPL_ID
    ) 
)

# save(fpl05, uds05, db01,
#      file = "07_db01.rdata")
# load("07_db01.rdata")
```

## 2022/2023's prices

The objective is to find prices for all players in the upcoming season. We will do another web-scrapping on the FPL official database.

```{r db_official}
(off00 <-
   fromJSON("https://fantasy.premierleague.com/api/bootstrap-static/"))
```

By exploring the JSON data set, we can obtain the prices.

```{r db_prices}
(
  pr01 <- tibble(off00$elements) %>%
    select(first_name,
           second_name,
           web_name,
           now_cost,
           team,
           team_code,
           # id,
           element_type,          
           # points_per_game,
           # selected_by_percent,total_points,
           # value_form,
           # value_season,
           # minutes,
           # goals_scored,
           # assists,
           # clean_sheets,
           # goals_conceded,
           # saves,
           # bps,
           # influence,
           # creativity,
           # threat,
           # ict_index,
           # influence_rank,
           # influence_rank_type,
           # creativity_rank,
           # creativity_rank_type,
           # threat_rank,
           # threat_rank_type,
           # ict_index_rank,
           # ict_index_rank_type
           ) %>%
    mutate(now_cost = now_cost / 10)
    )

# save(pr01,
#      file = "08_db_prices.rdata")
# load("08_db_prices.rdata")
```

...we also can get information about the strength of each team (difficulty rating).

```{r team}
#team
(tm01 <- tibble(off00$teams) %>%
  select(id, name, short_name, strength) %>% 
   arrange(desc(strength)))

# save(tm01,
#      file = "09_team.rdata")
# load("09_team.rdata")
```

...and the position for each player.

```{r position}
(
  pos01 <- tibble(off00$element_types) %>%
    select(id, singular_name, singular_name_short)
)

# save(pos01,
#      file = "10_position.rdata")
# load("10_position.rdata")
```

Now, we can merge all information from the official website.

```{r db02}
(
  db02 <- pr01 %>%
    left_join(tm01, by = c("team" = "id")) %>%
    left_join(pos01, by = c("element_type" = "id")) %>%
    rename(
      "team_name" = name,
      "position_name" = singular_name,
      "team_strength" = strength
    )  %>%
    #str()
    mutate_if(is.integer, as.numeric) %>%
    #str()
    mutate(full_name = paste(first_name, second_name, sep = " ")) %>%
    select(
      -team,
      -team_code,
      -element_type,
      -first_name,
      -second_name
    ) %>%
    relocate(full_name, team_name, team_strength, position_name, now_cost)
)

# save(db02,
#      file = "11_db02.rdata")
# load("11_db02.rdata")
```

```{r partial_match}
# NOTE: IGNORE THESE CODES

# (
#   db03 <- db02 %>%
#     fuzzy_full_join(
#       dict06,
#       by = c("full_name" = "Understat_Name", "singular_name_short" = "position"),
#       match_fun = stringi::stri_detect_fixed
#     )
# )

# xlsx::write.xlsx(db02, file = "20220708_matching.xlsx", sheetName = "new", 
#   col.names = TRUE, row.names = TRUE, append = FALSE)
# xlsx::write.xlsx(dict06, file = "20220708_matching.xlsx", sheetName = "old", 
#   col.names = TRUE, row.names = TRUE, append = TRUE)

# Check duplicated players
# (db04 <- db03 %>% group_by(full_name) %>% filter(n()>1))

```

...synchronising new players (previous seasons vs upcoming season) and taking care the latest transfers (up to 3 August 2022), which have not been included in the database.

```{r db04_before}
db03 <-
  "https://raw.githubusercontent.com/arifpras/BelutListrik/main/idlist2223.csv"

# db03 <- "~/OneDrive - The University of Nottingham/BB_SideProject/BelutListrik/idlist2223.csv"

(db04 <- read_csv(db03, show_col_types = FALSE))
```

```{r db02_vs_db04}
# (
#   db02_db04 <- db02 %>%
#     anti_join(db04, by = c("full_name")) %>%
#     left_join(dict06, by = c("full_name" = "name")) %>%
#     select(-Understat_ID,-web_name, -singular_name_short) %>%
#     mutate(name = ifelse(Understat_Name == "", "", full_name)) %>%
#     relocate(
#       "full_name",
#       "team_name",
#       "team_strength",
#       "position_name",
#       "now_cost",
#       "short_name",
#       "name",
#       "position",
#       "FPL_ID",
#       "Understat_Name"
#     )
# )

## With index
# xlsx::write.xlsx(db02_db04, file = "20220715_matching.xlsx", sheetName = "new",
#                  col.names = TRUE, row.names = TRUE, append = FALSE)

## No index
# xlsx::write.xlsx(
#   as.data.frame(db02_db04),
#   file = "20220803_matching.xlsx",
#   sheetName = "new",
#   col.names = TRUE,
#   row.names = FALSE,
#   append = FALSE,
#   showNA = FALSE
# )
```

Check if there are duplicates players (i.e., similar players but with different positions).

```{r db04_after}
# db03 <- "~/OneDrive - The University of Nottingham/BB_SideProject/BelutListrik/idlist2223.csv"
# 
# (db04 <- read_csv(db03, show_col_types = FALSE))

# (
#   db04 <- db02 %>%
#     bind_rows(db02_db04) %>%
#     arrange(full_name) %>% 
#     distinct(full_name, team_name, .keep_all = TRUE)
# )

db04 %>% 
  arrange(full_name) %>% 
  group_by(full_name, team_name) %>% filter(n() > 1) %>% 
  ungroup()

# save(db04,
#      file = "12_db04.rdata")
# load("12_db04.rdata")
```

Combining all players' statistics.

```{r db05}
(
  db05 <- db04 %>% select(-Understat_Name) %>%
    left_join(db01, by = c("FPL_ID"))
)

str(db05)

(
  db06 <- db05 %>%
    left_join(db02 %>% select(full_name, web_name, team_name), by = c("full_name", "team_name")) %>%
    select(-short_name,-name,-FPL_ID,-FPL_Name, -Understat_Name,-Understat_ID) %>%
    relocate(full_name, web_name)
)

# save(db05, db06,
#      file = "13_db05.rdata")
# load("13_db05.rdata")
```

## Fixture difficulty

First, we need to obtain the upcoming fixtures. Another web-screapping will be done on the following website <https://fixturedownload.com/results/epl-2022>.

```{r fixture}
# install an add-onn 'SelectorGadget' on Chrome

fixt00 <- "https://fixturedownload.com/results/epl-2022"
fixt01 <- read_html(fixt00)

fixt02 <- fixt01 %>%
  html_node("body") %>%
  html_children()

fixt03 <- html_nodes(fixt02, 'td')

fixt04 <- html_text(fixt03)
fixt05 <- data.frame(fixt04)

(fixt06 <-
    as.data.frame(matrix(
      fixt05$fixt04, ncol = 6, byrow = TRUE
    )))

# save(fixt01, fixt02, fixt03, fixt05, fixt06,
#      file = "14_fixt06.rdata")
# load("14_fixt06.rdata")
```

Then, we will combine the fixtures with each team's strength.

```{r strength}
(
  st01 <- fixt06 %>%
    select(V1, V4, V5) %>%
    rename(
      gameweek = V1,
      home = V4,
      away = V5
    ) %>%
    mutate(
      home = recode(home, "Nottingham Forest" = "Nott'm Forest"),
      away = recode(away, "Nottingham Forest" = "Nott'm Forest")
    ) %>%
    left_join(tm01, by = c("away" = "name")) %>%
    select(-id,-short_name) %>%
    rename(diff_away = strength)  %>% 
    # ~ how much the difficulty facing the away team?
    left_join(tm01, by = c("home" = "name")) %>%
    select(-id,-short_name) %>%
    rename(diff_home = strength)
  # ~ how much the difficulty facing the home team?
)

(
  st02 <- st01 %>%
    select(gameweek, home, diff_away) %>%
    rename(team = home,
           opponent_diff = diff_away)
)

(
  st03 <- st01 %>%
    select(gameweek, away, diff_home) %>%
    rename(team = away,
           opponent_diff = diff_home)
)

(st04 <- st02 %>%
    bind_rows(st03) %>%
    arrange(gameweek) %>% 
    mutate(
      gameweek = as.numeric(gameweek),
      opponent_diff = as.numeric(opponent_diff)
    )
)

# save(st01, st02, st03, st04,
#      file = "15_strength.rdata")
# load("15_strength.rdata")
```

Now, we want to aggregate level of difficulty in the next 8 matches (the World Cup will start after Gameweek 16).

```{r diff_8matches}
(diff01 <- st04 %>%
   arrange(team, gameweek) %>%
   filter(gameweek <= 8) %>%
   group_by(team) %>%
   summarise(diff_opponent = mean(opponent_diff)) %>%
   ungroup() %>% 
   arrange(diff_opponent))

# save(diff01,
#      file = "16_diff_8matches.rdata")
# load("16_diff_8matches.rdata")
```

Combining the next 8 matches difficulty to players' statistics.

```{r db07_final}
(
  db07 <- db06 %>%
    left_join(diff01, by = c ("team_name" = "team")) %>%
    mutate(player_id = row_number()) %>%
    relocate(player_id, full_name, web_name) %>%
    left_join(tm01 %>% select(name, short_name), by = c("team_name" = "name")) %>%
    mutate(player_team = paste(web_name, short_name, sep = " | ")) %>%
    relocate(
      player_id,
      player_team,
      position_name,
      now_cost,
      avg_points,
      avg_minutes,
      avg_ict,
      avg_influence,
      avg_xG,
      avg_xA,
      avg_xGChain,
      diff_opponent
    ) %>%
    mutate_if(is.numeric , replace_na, replace = 0)
)

# save(db07,
#      file = "17_db07_final.rdata")
# load("17_db07_final.rdata")

# xlsx::write.xlsx(
#   db07,
#   file = "20220803_fpl.xlsx",
#   sheetName = "db07",
#   col.names = TRUE,
#   row.names = TRUE,
#   append = FALSE
# )
```

# Building team

## Grouping by positions

```{r pos01}
(
  pos01 <- db07 %>% 
    select(player_team, position_name) %>% 
    as_tibble() %>% 
    arrange(position_name) %>% 
    group_by(position_name) %>% 
    mutate(
      pos_id = row_number()
    ) %>% 
    ungroup() %>% 
    pivot_wider(
      names_from = position_name,
      values_from = player_team
    ) %>% 
    select(-pos_id) %>% 
    relocate(
      Goalkeeper, Defender, Midfielder, Forward
    )
)

# save(pos01,
#      file = "18_pos01.rdata")
# load("17_db07_final.rdata")
```

How many combinations (without repetition) will we have?

$C_{k}(n) = \binom{n}{k} = \frac{n!}{k!(n-k)!}$, where

n = Elements to choose from (i.e., GK = 55 players available)

k = Elements chosen (i.e., maximum GK in the team = 2)

```{r comb_no_rep}
# References:
# https://www.hackmath.net/en/calculator/combinations-and-permutations?n=52&k=2&order=0&repeat=0
# https://rpruim.github.io/s341/S19/from-class/MathinRmd.html
# https://koalatea.io/r-combinations-permutations/

(
  gk_n <-
    db07 %>% filter(position_name %in% "Goalkeeper") %>% summarise(n = n()) %>% as.numeric()
)

(def_n <-
    db07 %>% filter(position_name %in% "Defender") %>% summarise(n = n())%>% as.numeric())

(mid_n <-
    db07 %>% filter(position_name %in% "Midfielder") %>% summarise(n = n())%>% as.numeric())

(fwd_n <-
    db07 %>% filter(position_name %in% "Forward") %>% summarise(n = n())%>% as.numeric())

gk_k <- as.numeric(2)
def_k <- as.numeric(5)
mid_k <- as.numeric(5)
fwd_k <- as.numeric(3)

(gk_choose <- choose(n = gk_n, k = gk_k))
(def_choose <- choose(n = def_n, k = def_k))
(mid_choose <- choose(n = mid_n, k = mid_k))
(fwd_choose <- choose(n = fwd_n, k = fwd_k))
```

Goalkeeper = `r toString(gk_choose)` Defender = `r toString(def_choose)` Midfielder = `r toString(mid_choose)` Forward = `r toString(fwd_choose)`

## GK: A group with 2 players

First we want to make a group consists of 2 players.

```{r gk02}
tic("Time needed:") # 0.023 sec elapsed
set.seed(19) #random number

gk01 <- db07 %>% filter(position_name %in% "Goalkeeper")
gk02 <-  t(combn(gk01$player_team, 2)) #max: 2 players

toc()
beepr::beep(sound = 1)

# save(gk02,
#      file = "18_gk02.rdata")
# load("18_gk02.rdata")
```

...then combine with their statistics.

```{r gk03}
tic("Time needed:") # 0.099 sec elapsed

(
  gk03 <- gk02 %>%
    # na.omit() %>%
    as_tibble() %>%
    mutate("id_gk" = row_number()) %>%
    relocate(id_gk) %>%
    pivot_longer(
      cols = 2:3,
      names_to = "v",
      values_to = "player_team"
    ) %>%
    select(-v) %>%
    left_join(db07, by = "player_team")
)

toc()
beepr::beep(sound = 1)
```

...aggregate statistics for each group.

```{r gk04}
tic("Time needed:") #0.089 sec elapsed

(
  gk04 <- gk03 %>%
    group_by(id_gk) %>%
    # filter(any(player_team == "Mendy | CHE")) %>% #Mendy is a must
    summarise(
      sum_price = sum(now_cost, na.rm = FALSE),
      #sum, because we have maximal budget
      #avg_avgpoints = mean(avg_points, na.rm = FALSE),
      sum_avgpoints = mean(avg_points, na.rm = FALSE), #sum for mean, so can be compared
      #1 main, 1 bench
      #avg_avgpoints = mean(avg_points, na.rm = FALSE), #both could be main
      #average, because we have to pick 1 players
      avg_avgminutes = mean(avg_minutes, na.rm = FALSE),
      avg_avgict = mean(avg_ict, na.rm = FALSE),
      avg_avginfluence = mean(avg_influence, na.rm = FALSE),
      avg_avgXG = mean(avg_xG, na.rm = FALSE),
      avg_avgxA = mean(avg_xA, na.rm = FALSE),
      avg_avgxGChain = mean(avg_xGChain, na.rm = FALSE),
      avg_diffopp = mean(diff_opponent, na.rm = FALSE),
    ) %>%
    arrange(desc(sum_avgpoints),
            sum_price) %>%
    ungroup()
)

toc()
beepr::beep(sound = 1)
```

Let's say the budget for the goalkeeper position is £9.0.

```{r gk05}
tic("Time needed:") #0.028 sec elapsed

(
  gk05 <- gk04 %>%
    arrange(desc(sum_avgpoints)) %>%
    filter(sum_price <= 9) %>% #restrict the price
    slice_head(n = 5) %>% #the closest to 10
    left_join(gk03, by = "id_gk") %>% 
    relocate(id_gk, player_team)
)

toc()
beepr::beep(sound = 1)
```

...it's going to be easier to plot the data.

```{r gk_plot}
  (
    gk_plot01 <- gk05 %>%
      select(
        id_gk,
        sum_price,
        sum_avgpoints,
        avg_avgminutes,
        avg_avgict,
        avg_avginfluence,
        avg_avgXG,
        avg_avgxA,
        avg_avgxGChain,
        avg_diffopp
      ) %>%
      group_by(id_gk) %>%
      slice_head(n = 1) %>%
      ungroup() %>%
      pivot_longer(
        cols = 2:10,
        names_to = "indicator",
        values_to = "value"
      ) %>%
      ggplot(aes(
        x = reorder_within(id_gk, value, indicator),
        y = value,
        fill = factor(id_gk)
      )) +
      geom_col() +
      #geom_bar(stat = "identity", aes(fill = factor(id_gk))) +
      coord_flip() +
      scale_x_reordered() +
      #scale_fill_viridis(discrete = TRUE) +
      #scale_fill_manual(values = wes_palette("Darjeeling1")) +
      scale_fill_manual(values = c(
        "#ef476f", "#ffd166", "#06d6a0", "#118ab2", "#073b4c"
      )) +
      egg::theme_article() +
      facet_wrap(~ indicator, ncol = 3, scales = "free") +
      theme(legend.position = "none") +
      labs(x = "",
           y = "")
)
```

Which GK team?

```{r gk_choice}
gk_team <- "943"

(gk_choice <-
    gk05 %>% filter(id_gk == gk_team) %>% mutate("position" = "GK"))
```

Take a loot at the player's statistics individually.

```{r gk_indv_stat}
(
  gk_plot02 <- gk_choice %>%
    select(
      player_team,
      now_cost,
      avg_points,
      avg_minutes,
      avg_ict,
      avg_influence,
      avg_xG,
      avg_xA,
      avg_xGChain,
      diff_opponent
    ) %>%
    pivot_longer(
      cols = 2:10,
      names_to = "indicator",
      values_to = "value"
    ) %>%
    ggplot(aes(
      x = reorder_within(player_team, value, indicator),
      y = value,
      fill = factor(player_team)
    )) +
    geom_col() +
    #geom_bar(stat = "identity", aes(fill = factor(id_gk))) +
    coord_flip() +
    scale_x_reordered() +
    #scale_fill_viridis(discrete = TRUE) +
    #scale_fill_manual(values = wes_palette("Darjeeling1")) +
    scale_fill_manual(values = c("#6a4c93", "#1982c4")) +
    egg::theme_article() +
    facet_wrap( ~ indicator, ncol = 3, scales = "free") +
    theme(legend.position = "none") +
    labs(x = "",
         y = "")
)
```

## DEF: A group with 5 players

A group consists of 5 players.

```{r def02}
set.seed(19) #random number

def01 <- db07 %>% filter(position_name %in% "Defender") %>%
  arrange(desc(now_cost))

def_high <- def01 %>% slice_head(prop = 0.20)
def_low <- def01 %>% slice_tail(prop = 0.20)
def_comb <- def_high %>%
  bind_rows(def_low)

choose(n = def_comb %>% summarise(n = n()) %>% as.numeric(), k = 5)

tic("Time needed:") # 15.571 sec elapsed
def02 <- t(combn(def_comb$player_team, 5)) #max: 5 players
toc()
beepr::beep(sound = 1)

# save(def02,
#      file = "19_def02.rdata")
# load("19_def02.rdata")
```

...then combine with their statistics.

```{r def03}
tic("Time needed:") # 67.503 sec elapsed
(
  def03 <- def02 %>%
    # na.omit() %>%
    as_tibble() %>%
    mutate("id_def" = row_number()) %>%
    relocate(id_def) %>%
    pivot_longer(
      cols = 2:6,
      names_to = "v",
      values_to = "player_team"
    ) %>%
    select(-v) %>%
    left_join(db07, by = "player_team")
)
toc()
beepr::beep(sound = 1)
```

```{r}
# save(def03,
#      file = "20_def03.rdata")
# load("20_def03.rdata")

# saveRDS(def03,
#         file = "20_def03.rds")

# tic("Time needed:") # 201.467 sec elapsed
# def03 <- readRDS("20_def03.rds")
# toc()
```

...aggregate statistics for each group.

```{r def04}
tic("Time needed:") # 306.567 sec elapsed
(
  def04 <- def03 %>%
    group_by(id_def) %>%
    filter(any(
      player_team %in% c("N.Williams | NFO") 
      # I want him in my squad, use "any!()" if you want any particular not included in your squad
    )) %>%
    summarise(
      sum_price = sum(now_cost),
      #sum, because we have maximal budget
      sum_avgpoints = sum(avg_points),
      #sum, because we can pick 5 players
      avg_avgminutes = mean(avg_minutes),
      avg_avgict = mean(avg_ict),
      avg_avginfluence = mean(avg_influence),
      avg_avgXG = mean(avg_xG),
      avg_avgxA = mean(avg_xA),
      avg_avgxGChain = mean(avg_xGChain),
      avg_diffopp = mean(diff_opponent),
    ) %>%
    arrange(desc(sum_avgpoints),
            sum_price) %>% 
    ungroup()
)
toc()
beepr::beep(1)
```

```{r}
# save(def04,
#      file = "21_def04.rdata")

# tic("Time needed:") #0.318 sec elapsed
# load("~/OneDrive - The University of Nottingham/BB_SideProject/BelutListrik/21_def04.rdata")
# toc()

# saveRDS(def04,
#         file = "21_def04.rds")

# tic("Time needed:")
# def04 <- readRDS("21_def04.rds")
# toc()
```

Let's say the budget for the defender position is £35.0.

```{r def05}
tic("Time needed:") # 20.069 sec elapsed

(
  def05 <- def04 %>%
    arrange(desc(sum_avgpoints)) %>%
    filter(sum_price <= 35 & sum_price > 20) %>%
    slice_head(n = 5) %>%
    left_join(def03, by = "id_def") %>%
    relocate(id_def, player_team)
)

toc()
beepr::beep(sound = 1)
```

```{r}
# saveRDS(def05,
#         file = "22_def05.rds")

# tic("Time needed:")
# def05 <- readRDS("/~/OneDrive - The University of Nottingham/BB_SideProject/BelutListrik/22_def05.rds")
# toc()
```

...it's going to be easier to plot the data.

```{r def_plot}
tic("Time needed:") #2.157 sec elapsed

(
  def_plot01 <- def05 %>%
    select(
      id_def,
      sum_price,
      sum_avgpoints,
      avg_avgminutes,
      avg_avgict,
      avg_avginfluence,
      avg_avgXG,
      avg_avgxA,
      avg_avgxGChain,
      avg_diffopp
    ) %>%
    group_by(id_def) %>%
    slice_head(n = 1) %>%
    ungroup() %>%
    pivot_longer(
      cols = 2:10,
      names_to = "indicator",
      values_to = "value"
    ) %>%
    ggplot(aes(
      x = reorder_within(id_def, value, indicator),
      y = value,
      fill = factor(id_def)
    )) +
    geom_col() +
    #geom_bar(stat = "identity", aes(fill = factor(id_gk))) +
    coord_flip() +
    scale_x_reordered() +
    #scale_fill_viridis(discrete = TRUE) +
    #scale_fill_manual(values = wes_palette("Darjeeling1")) +
    # scale_fill_manual(values = c(
    #   "#ef476f", "#ffd166", "#06d6a0", "#118ab2", "#073b4c"
    # )) +
    scale_fill_manual(values = c(
      "#6a4c93", "#1982c4", "#8ac926", "#ffca3a", "#ff595e"
    )) +
    egg::theme_article() +
    facet_wrap( ~ indicator, ncol = 3, scales = "free") +
    theme(legend.position = "none") +
    labs(x = "",
         y = "")
)

toc()
beepr::beep(sound = 1)
```

Which DEF team?

```{r def_choice}
def_team <- "434"

(def_choice <-
    def05 %>% filter(id_def == def_team) %>% mutate("position" = "DEF"))
```

Take a look at the player's statistics individually. In addition, it turns out, £31.5 is enough. Our budget was £35.0. We will add £3.5 to the midfielders' budget. So the budget for midfielders will be £38.5, instead of £35.0.

```{r def_indv_stat}
(
  def_plot02 <- def_choice %>%
    select(
      player_team,
      now_cost,
      avg_points,
      avg_minutes,
      avg_ict,
      avg_influence,
      avg_xG,
      avg_xA,
      avg_xGChain,
      diff_opponent
    ) %>%
    pivot_longer(
      cols = 2:10,
      names_to = "indicator",
      values_to = "value"
    ) %>%
    ggplot(aes(
      x = reorder_within(player_team, value, indicator),
      y = value,
      fill = factor(player_team)
    )) +
    geom_col() +
    #geom_bar(stat = "identity", aes(fill = factor(id_gk))) +
    coord_flip() +
    scale_x_reordered() +
    #scale_fill_viridis(discrete = TRUE) +
    #scale_fill_manual(values = wes_palette("Darjeeling1")) +
    scale_fill_manual(values = c("#6a4c93", "#1982c4", "#8ac926", "#ffca3a", "#ff595e")) +
    egg::theme_article() +
    facet_wrap( ~ indicator, ncol = 3, scales = "free") +
    theme(legend.position = "none") +
    labs(x = "",
         y = "")
)
```

## MID: A group with 5 players

A group consists of 5 players.

```{r mid02}
set.seed(19) #random number

mid01 <- db07 %>% filter(position_name %in% "Midfielder") %>%
  arrange(desc(now_cost))

mid_high <- mid01 %>% slice_head(prop = 0.25)
mid_low <- mid01 %>% slice_tail(prop = 0.15)
mid_comb <- mid_high %>%
  bind_rows(mid_low)

choose(n = mid_comb %>% summarise(n = n()) %>% as.numeric(), k = 5)

tic("Time needed:") # 50.396 sec elapsed
mid02 <- t(combn(mid_comb$player_team, 5)) #max: 5 players
toc()
beepr::beep(sound = 1)
```

```{r}
# save(mid02,
#      file = "23_mid02.rdata")

# tic("Time needed:") # 49.442 sec elapsed
# load("~/OneDrive - The University of Nottingham/BB_SideProject/BelutListrik/23_mid02.rdata")
# toc()
```

...then combine with their statistics.

```{r mid03}
tic("Time needed:") # 223.617 sec elapsed
(
  mid03 <- mid02 %>%
    # na.omit() %>%
    as_tibble() %>%
    mutate("id_mid" = row_number()) %>%
    relocate(id_mid) %>%
    pivot_longer(
      cols = 2:6,
      names_to = "v",
      values_to = "player_team"
    ) %>%
    select(-v) %>%
    left_join(db07, by = "player_team")
)
toc()
beepr::beep(sound = 1)
```

```{r}
# save(mid03,
#      file = "24_mid03.rdata")

# tic("Time needed:") # 471.272 sec elapsed
# load("~/OneDrive - The University of Nottingham/BB_SideProject/BelutListrik/24_mid03.rdata")
# toc()

# saveRDS(mid03,
#         file = "24_mid03.rds")
# mid03 <- readRDS("24_mid03.rds")
```

...aggregate statistics for each group.

```{r mid04}
tic("Time needed:") #

(
  mid04 <- mid03 %>%
    group_by(id_mid) %>%
    filter(any(player_team %in% c("Salah | LIV"))) %>%
    filter(!any(player_team %in% c("Raphinha | LEE"))) %>%
    summarise(
      sum_price = sum(now_cost, na.rm = FALSE),
      #sum, because we have maximal budget
      sum_avgpoints = sum(avg_points, na.rm = FALSE),
      #sum, because we can pick 5 players
      avg_avgminutes = mean(avg_minutes, na.rm = FALSE),
      avg_avgict = mean(avg_ict, na.rm = FALSE),
      avg_avginfluence = mean(avg_influence, na.rm = FALSE),
      avg_avgXG = mean(avg_xG, na.rm = FALSE),
      avg_avgxA = mean(avg_xA, na.rm = FALSE),
      avg_avgxGChain = mean(avg_xGChain, na.rm = FALSE),
      avg_diffopp = mean(diff_opponent, na.rm = FALSE),
    ) %>%
    arrange(desc(sum_avgpoints),
            sum_price) %>%
    ungroup()
)

toc()
beepr::beep(1)
```

```{r mid04}
# tic("Time needed:") #
# 
# (
#   mid04_kulu <- mid03 %>%
#     group_by(id_mid) %>%
#     filter(any(player_team %in% c("Salah | LIV"))) %>%
#     filter(any(player_team %in% c("Kulusevski | TOT"))) %>%
#     filter(!any(player_team %in% c("Raphinha | LEE"))) %>%
#     summarise(
#       sum_price = sum(now_cost, na.rm = FALSE),
#       #sum, because we have maximal budget
#       sum_avgpoints = sum(avg_points, na.rm = FALSE),
#       #sum, because we can pick 5 players
#       avg_avgminutes = mean(avg_minutes, na.rm = FALSE),
#       avg_avgict = mean(avg_ict, na.rm = FALSE),
#       avg_avginfluence = mean(avg_influence, na.rm = FALSE),
#       avg_avgXG = mean(avg_xG, na.rm = FALSE),
#       avg_avgxA = mean(avg_xA, na.rm = FALSE),
#       avg_avgxGChain = mean(avg_xGChain, na.rm = FALSE),
#       avg_diffopp = mean(diff_opponent, na.rm = FALSE),
#     ) %>%
#     arrange(desc(sum_avgpoints),
#             sum_price) %>%
#     ungroup()
# )
# 
# toc()
# beepr::beep(1)
```

```{r}
# tic("Time needed:") #
# 
# (
#   mid04_all <- mid03 %>%
#     group_by(id_mid) %>%
#     # filter(any(player_team %in% c("Salah | LIV"))) %>%
#     # filter(any(player_team %in% c("Kulusevski | TOT"))) %>%
#     filter(!any(player_team %in% c("Raphinha | LEE"))) %>%
#     summarise(
#       sum_price = sum(now_cost, na.rm = FALSE),
#       #sum, because we have maximal budget
#       sum_avgpoints = sum(avg_points, na.rm = FALSE),
#       #sum, because we can pick 5 players
#       avg_avgminutes = mean(avg_minutes, na.rm = FALSE),
#       avg_avgict = mean(avg_ict, na.rm = FALSE),
#       avg_avginfluence = mean(avg_influence, na.rm = FALSE),
#       avg_avgXG = mean(avg_xG, na.rm = FALSE),
#       avg_avgxA = mean(avg_xA, na.rm = FALSE),
#       avg_avgxGChain = mean(avg_xGChain, na.rm = FALSE),
#       avg_diffopp = mean(diff_opponent, na.rm = FALSE),
#     ) %>%
#     arrange(desc(sum_avgpoints),
#             sum_price) %>%
#     ungroup()
# )
# 
# toc()
# beepr::beep(1)
```

```{r}
# saveRDS(mid04,
#         file = "25_mid04.rds")

# saveRDS(mid04,
#         file = "25_mid04_kulu.rds")

# tic("Time needed:") # 0.508 sec elapsed
# mid04 <- readRDS("~/OneDrive - The University of Nottingham/BB_SideProject/BelutListrik/25_mid04.rds")
# toc()
```

Let's say the budget for the midfielder position is £38.5.

```{r mid05}
tic("Time needed:") #119.773 sec elapsed
(
  mid05 <- mid04 %>%
    arrange(desc(sum_avgpoints)) %>%
    filter(sum_price <= 38.5 & sum_price > 30) %>%
    slice_head(n = 5) %>%
    left_join(mid03, by = "id_mid") %>%
    relocate(id_mid, player_team)
)
toc()
beepr::beep(1)
```

```{r}
# saveRDS(mid05,
#         file = "26_mid05.rds")

# tic("Time needed:") #
# mid05 <- readRDS("26_mid05.rds")
# toc()
```

...it's going to be easier to plot the data.

```{r mid_plot}
(
  mid_plot01 <- mid05 %>%
    select(
      id_mid,
      sum_price,
      sum_avgpoints,
      avg_avgminutes,
      avg_avgict,
      avg_avginfluence,
      avg_avgXG,
      avg_avgxA,
      avg_avgxGChain,
      avg_diffopp
    ) %>%
    group_by(id_mid) %>%
    slice_head(n = 1) %>%
    ungroup() %>%
    pivot_longer(
      cols = 2:10,
      names_to = "indicator",
      values_to = "value"
    ) %>%
    ggplot(aes(
      x = reorder_within(id_mid, value, indicator),
      y = value,
      fill = factor(id_mid)
    )) +
    geom_col() +
    #geom_bar(stat = "identity", aes(fill = factor(id_gk))) +
    coord_flip() +
    scale_x_reordered() +
    #scale_fill_viridis(discrete = TRUE) +
    #scale_fill_manual(values = wes_palette("Darjeeling1")) +
    scale_fill_manual(values = c(
      "#ef476f", "#ffd166", "#06d6a0", "#118ab2", "#073b4c"
    )) +
    egg::theme_article() +
    facet_wrap(~ indicator, ncol = 3, scales = "free") +
    theme(legend.position = "none") +
    labs(x = "",
         y = "")
)
```

Which MID team?

```{r mid_choice}
mid_team <- "753102"

(mid_choice <-
    mid05 %>% filter(id_mid == mid_team) %>% mutate("position" = "MID"))
```

Take a loot at the player's statistics individually.

```{r mid_indv_stat}
(
  mid_plot02 <- mid_choice %>%
    select(
      player_team,
      now_cost,
      avg_points,
      avg_minutes,
      avg_ict,
      avg_influence,
      avg_xG,
      avg_xA,
      avg_xGChain,
      diff_opponent
    ) %>%
    pivot_longer(
      cols = 2:10,
      names_to = "indicator",
      values_to = "value"
    ) %>%
    ggplot(aes(
      x = reorder_within(player_team, value, indicator),
      y = value,
      fill = factor(player_team)
    )) +
    geom_col() +
    #geom_bar(stat = "identity", aes(fill = factor(id_gk))) +
    coord_flip() +
    scale_x_reordered() +
    #scale_fill_viridis(discrete = TRUE) +
    #scale_fill_manual(values = wes_palette("Darjeeling1")) +
    scale_fill_manual(values = c("#6a4c93", "#1982c4", "#8ac926", "#ffca3a", "#ff595e")) +
    egg::theme_article() +
    facet_wrap( ~ indicator, ncol = 3, scales = "free") +
    theme(legend.position = "none") +
    labs(x = "",
         y = "")
)
```

## FWD: A group with 3 players

A group consists of 3 players.

```{r fwd02}
set.seed(19) #random number

fwd01 <- db07 %>% filter(position_name %in% "Forward") %>%
  arrange(desc(now_cost))

# fwd_high <- fwd01 %>% slice_head(prop = 0.2)
# fwd_low <- fwd01 %>% slice_tail(prop = 0.2)
# fwd_comb <- fwd_high %>%
#   bind_rows(fwd_low)

choose(n = fwd01 %>% summarise(n = n()) %>% as.numeric(), k = 3)

tic("Time needed:") #0.023 sec elapsed
fwd02 <- t(combn(fwd01$player_team, 3)) #max: 3 players
toc()
beepr::beep(sound = 1)
```

...then combine with their statistics.

```{r fwd03}
tic("Time needed:") #0.105 sec elapsed
(
  fwd03 <- fwd02 %>%
    # na.omit() %>%
    as_tibble() %>%
    mutate("id_fwd" = row_number()) %>%
    relocate(id_fwd) %>%
    pivot_longer(
      cols = 2:4,
      names_to = "v",
      values_to = "player_team"
    ) %>%
    select(-v) %>%
    left_join(db07, by = "player_team")
)
toc()
beepr::beep(sound = 1)
```

...aggregate statistics for each group.

```{r fwd04}
tic("Time needed:") #0.284 sec elapsed

(
  fwd04 <- fwd03 %>%
    group_by(id_fwd) %>%
    #filter(any(player_team %in% c("Haaland | MCI"))) %>%  
    filter(any(player_team %in% c("Jesus | ARS"))) %>% 
    summarise(
      sum_price = sum(now_cost, na.rm = FALSE),
      #sum, because we have maximal budget
      sum_avgpoints = sum(avg_points, na.rm = FALSE),
      #sum, because we can pick 5 players
      avg_avgminutes = mean(avg_minutes, na.rm = FALSE),
      avg_avgict = mean(avg_ict, na.rm = FALSE),
      avg_avginfluence = mean(avg_influence, na.rm = FALSE),
      avg_avgXG = mean(avg_xG, na.rm = FALSE),
      avg_avgxA = mean(avg_xA, na.rm = FALSE),
      avg_avgxGChain = mean(avg_xGChain, na.rm = FALSE),
      avg_diffopp = mean(diff_opponent, na.rm = FALSE),
    ) %>%
    arrange(desc(sum_avgpoints),
            sum_price) %>%
    ungroup()
)

# saveRDS(fwd04, "28_fwd04.rds")
toc()
beepr::beep(1)

```

Let's say the budget for the forwards is 21.

```{r fwd05}
tic("Time needed:") #0.04 sec elapsed
(
  fwd05 <- fwd04 %>%
    arrange(desc(sum_avgpoints)) %>%
    filter(sum_price <= 21) %>%
    slice_head(n = 5) %>%
    left_join(fwd03, by = "id_fwd") %>%
    relocate(id_fwd, player_team)
)
toc()
beepr::beep(1)

# saveRDS(fwd05, "29_fwd05.rds")
# fwd05 <- readRDS("~/OneDrive - The University of Nottingham/BB_SideProject/BelutListrik/29_fwd05.rds")
```

...it's going to be easier to plot the data.

```{r fwd_plot}
(
  fwd_plot01 <- fwd05 %>%
    select(
      id_fwd,
      sum_price,
      sum_avgpoints,
      avg_avgminutes,
      avg_avgict,
      avg_avginfluence,
      avg_avgXG,
      avg_avgxA,
      avg_avgxGChain,
      avg_diffopp
    ) %>%
    group_by(id_fwd) %>%
    slice_head(n = 1) %>%
    ungroup() %>%
    pivot_longer(
      cols = 2:10,
      names_to = "indicator",
      values_to = "value"
    ) %>%
    ggplot(aes(
      x = reorder_within(id_fwd, value, indicator),
      y = value,
      fill = factor(id_fwd)
    )) +
    geom_col() +
    #geom_bar(stat = "identity", aes(fill = factor(id_gk))) +
    coord_flip() +
    scale_x_reordered() +
    #scale_fill_viridis(discrete = TRUE) +
    #scale_fill_manual(values = wes_palette("Darjeeling1")) +
    scale_fill_manual(values = c("#6a4c93", "#1982c4", "#8ac926", "#ffca3a", "#ff595e")) +
    egg::theme_article() +
    facet_wrap(~ indicator, ncol = 3, scales = "free") +
    theme(legend.position = "none") +
    labs(x = "",
         y = "")
)
```

Which FWD team?

```{r fwd_choice}
fwd_team <- "12586"

(fwd_choice <-
    fwd05 %>% filter(id_fwd == fwd_team) %>% mutate("position" = "FWD"))
```

Take a loot at the player's statistics individually.

```{r fwd_indv_stat}
(
  fwd_plot02 <- fwd_choice %>%
    select(
      player_team,
      now_cost,
      avg_points,
      avg_minutes,
      avg_ict,
      avg_influence,
      avg_xG,
      avg_xA,
      avg_xGChain,
      diff_opponent
    ) %>%
    pivot_longer(
      cols = 2:10,
      names_to = "indicator",
      values_to = "value"
    ) %>%
    ggplot(aes(
      x = reorder_within(player_team, value, indicator),
      y = value,
      fill = factor(player_team)
    )) +
    geom_col() +
    #geom_bar(stat = "identity", aes(fill = factor(id_gk))) +
    coord_flip() +
    scale_x_reordered() +
    #scale_fill_viridis(discrete = TRUE) +
    #scale_fill_manual(values = wes_palette("Darjeeling1")) +
    scale_fill_manual(values = c("#6a4c93", "#1982c4", "#8ac926")) +
    egg::theme_article() +
    facet_wrap( ~ indicator, ncol = 3, scales = "free") +
    theme(legend.position = "none") +
    labs(x = "",
         y = "")
)
```

# Final formation

Combining all the positions.

```{r squad}
(belutlistrik_squad <- (gk_choice %>% rename("id" = id_gk)) %>%
   bind_rows((def_choice %>% rename("id" = id_def)),
             (mid_choice %>% rename("id" = id_mid)),
             (fwd_choice %>% rename("id" = id_fwd))
   ) %>% 
   mutate(
     "formid" = row_number()
   ) %>% 
   relocate(id, formid, player_team, position) %>% 
   mutate(
     player_team = recode(player_team, "Alexander-Arnold | LIV" = "Trent | LIV")
   )
 )
```

Checking the total price to make sure it less than 100.

```{r sum_price}
belutlistrik_squad %>% select(player_team, position_name, now_cost) %>%
  summarise(across(where(is.numeric), sum))
```

Finding the most ideal formation by taking a look at the best first-10 in each indicator. For example, if the first-10 (from top to bottom) consists of 4 defenders, 4 midfielders, and 2 forwards maybe we want to choose 4-4-2.

```{r blselect}
(
  belutlistrik_select <- belutlistrik_squad %>%
    filter(position != "GK") %>%
    select(
      player_team,
      position,
      now_cost,
      avg_points,
      avg_minutes,
      avg_ict,
      avg_influence,
      avg_xG,
      avg_xA,
      avg_xGChain,
      diff_opponent
    ) %>%
    pivot_longer(
      cols = 3:10,
      names_to = "indicator",
      values_to = "value"
    ) %>%
    #mutate(
    #  "indicator" = as.factor(indicator),
    #  "Understat_Name" = tidytext::reorder_within(Understat_Name, value, indicator)
    #    ) %>%
    ggplot(aes(
      x = reorder_within(player_team, value, indicator),
      y = value,
      fill = factor(position)
    )) +
    #geom_col(show.legend = FALSE) +
    geom_col() +
    #geom_bar(stat = "identity", aes(fill = factor(position))) +
    coord_flip() +
    scale_x_reordered() +
    #scale_fill_viridis(discrete = TRUE) +
    #scale_fill_manual(values = wes_palette("Darjeeling1")) +
    scale_fill_manual(values = c("#ef476f", "#ffd166", "#06d6a0")) +
    egg::theme_article() +
    facet_wrap( ~ indicator, ncol = 4, scales = "free") +
    theme(legend.position = "top",
          legend.title = element_blank()) +
    labs(x = "",
         y = "")
  #guides(fill = guide_legend("Position:"))
)
```

...then we need to decide who are the main players. Look the order from top to bottom in `belutlistrik_squad`.

```{r bloptions}
(
  belutlistrik_options <- belutlistrik_squad %>%
    mutate(
      "form352" = ifelse(formid %in% c(2, 6, 7, 15), "bench", "play"),
      "form442" = ifelse(formid %in% c(2, 7, 12, 15), "bench", "play"),
      "form343" = ifelse(formid %in% c(2, 6, 7, 12), "bench", "play"),
      "form451" = ifelse(formid %in% c(2, 7, 14, 15), "bench", "play"),
    ) %>%
    relocate(
      id, formid, form442, form343, form352, form451, player_team
    )
)
```

Building the formations based on the coordinate of x and y in the plot.

```{r form_choice}
(for352 <- tibble(
  x = c(98, 98, 83, 83, 83, 89 ,80, 68, 68, 68, 68, 68, 53, 53, 71),
  y = c(50, 110, 25, 50, 75, 110, 110, 10, 30, 50, 70, 90, 37.5, 62.5, 110),
  formid = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
))

(for442 <- tibble(
  x = c(98, 98, 83, 83, 83, 83, 89 ,68, 68, 68, 68, 80, 53, 53, 71),
  y = c(50, 110, 20, 40, 60, 80, 110, 20, 40, 60, 80, 110, 37.5, 62.5, 110),
  formid = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
))

(for343 <- tibble(
  x = c(98, 98, 83, 83, 83, 89, 80, 68, 68, 68, 68, 71, 53, 53, 53),
  y = c(50, 110, 25, 50, 75, 110, 110, 20, 40, 60, 80, 110, 25, 50, 75),
  formid = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
))

(for451 <- tibble(
  x = c(98, 98, 83, 83, 83, 83, 89, 68, 68, 68, 68, 68, 53, 80, 71),
  y = c(50, 110, 20, 40, 60, 80, 110, 10, 30, 50, 70, 90, 50, 110, 110),
  formid = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
))
```

... assigning the choosen players to the formation.

```{r blform}
(
  belutlistrik_442 <- belutlistrik_options %>%
    select(formid, form442, player_team) %>%
    right_join(for442, by = "formid")
)

(
  belutlistrik_343 <- belutlistrik_options %>%
    select(formid, form343, player_team) %>%
    right_join(for343, by = "formid")
)

(
  belutlistrik_352 <- belutlistrik_options %>%
    select(formid, form352, player_team) %>%
    right_join(for352, by = "formid")
)

(
  belutlistrik_451 <- belutlistrik_options %>%
    select(formid, form451, player_team) %>%
    right_join(for451, by = "formid")
)
```

Plotting the final formation.

```{r blfinal}
(
  belutlistrik <- ggplot(belutlistrik_442) + #change
    annotate_pitch(
      colour = "white",
      fill   = "#e0e2db",
      limits = FALSE
    ) +
    geom_point(
      aes(x = x, y = 100 - y, fill = factor(form442)), #change
      shape = 21,
      size = 5
    ) +
    scale_fill_manual(values = c("play" = "#8b2635", "bench" = "#2e3532")) +
    geom_text_repel(
      aes(
        x = x,
        y = 100 - y,
        #label = stringr::str_wrap(player_team, 10),
        label = player_team,
        color = factor(form442) #change
      ),
      direction = "y",
      nudge_x = -4,
      segment.colour = NA,
      size = 3,
      #box.padding = unit(-2.5, "lines")
    ) +
    scale_color_manual(values = c("play" = "#8b2635", "bench" = "#2e3532")) +
    theme_pitch() +
    theme(
      panel.background = element_rect(fill = "#e0e2db"),
      legend.position = "none",
      plot.title = element_text(hjust = 0, size = 18,
                                face = "bold"),
      plot.subtitle = element_text(hjust = 0, size = 14),
    ) +
    coord_flip(xlim = c(49, 101),
               ylim = c(-12, 112)) +
    ggtitle("Belut Listrik FC",
            "instagram/arifpras")
)

#belutlistrik
```

# Clearing the environment

```{r clearing}
# Removes all items in Environment!
# rm(list = ls())
# ls()
```