Gole w europejskich ligach

# Będziemy potrzebowac standardowo pakietów:
library(tidyverse)
library(janitor)
library(igraph)

# a żeby pobrac dane z Google Drive wystarczy:
library(googledrive)

"1yeIl0rZldsmmRzdGUkG56QmX7UXI0PHQ" %>%
  as_id() %>%
  drive_get() %>%
  drive_download(path = "xg_master.csv")

data <- read_csv("xg_master.csv") %>%
  clean_names() %>%
  mutate(player = stringi::stri_enc_tonative(player),
         assist_player = stringi::stri_enc_tonative(assist_player))

data %>%
  distinct(league, team) %>%
  group_by(league) %>%
  arrange(team) %>%
  mutate(n = row_number()) %>%
  ungroup() %>%
  spread(league, team, fill = "") %>%
  select(-n) %>%
  kable() %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

# skąd co pada?
data %>%
  ggplot() +
  geom_point(aes(x_coordinate, y_coordinate,
                 color = situation),
             alpha = 0.7, size = 0.5, show.legend = FALSE) +
  facet_grid(situation~result) +
  # boisko
  xlim(c(0, 1)) +
  ylim(c(0, 1)) +
  geom_rect(aes(xmin = 0, ymin = 0, xmax = 1, ymax = 1), fill = NA, color = "black") +
  geom_rect(aes(xmin = 0, ymin = 0, xmax = 0.5, ymax = 1), fill = NA, color = "black") +
  geom_rect(aes(xmin = 0, ymin = 0.2, xmax = 0.157, ymax = 0.8), fill = NA, color = "black") +
  geom_rect(aes(xmin = 1, ymin = 0.2, xmax = 1-0.157, ymax = 0.8), fill = NA, color = "black") +
  labs(title = "Skąd padają strzały?", x = "", y = "") +
  theme_void()

data %>% count(result, sort = TRUE) %>%
  kable() %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

data %>% filter(result == "Goal") %>% count(shot_type, sort = TRUE) %>%
  kable() %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

data %>% filter(result == "Goal") %>% count(situation, sort = TRUE) %>%
  kable() %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

data %>% filter(result == "Goal") %>% count(preceding_action, sort = TRUE) %>%
  kable() %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

data %>%
  filter(result == "Goal") %>%
  filter(minute <= 92) %>%
  count(minute) %>%
  ggplot() +
  geom_line(aes(minute, n), color = "green") +
  geom_smooth(aes(minute, n), color = "lightgreen", alpha = 0.5) +
  geom_vline(xintercept = c(45, 90)) +
  labs(title = "Liczba bramek w zależności od minuty meczu",
       x = "Minuta meczu",
       y = "Łączna liczba bramek")

# bloki po 5 minut, zależność od ligi
data %>%
  filter(result == "Goal") %>%
  mutate(minute = 5*( minute %/% 5) ) %>%
  count(minute, league) %>%
  group_by(league) %>%
  mutate(n = 100*n/sum(n)) %>%
  ungroup %>%
  ggplot() +
  geom_tile(aes(minute, league, fill = n), color = "gray50") +
  scale_fill_distiller(palette = "Reds", direction = 1) +
  geom_vline(xintercept = c(45, 90)) +
  labs(title = "Procent bramek w meczu zależności od minuty meczu",
       x = "Minuta meczu",
       y = "",
       fill = "Procent bramek w lidze")

data %>%
  filter(player == "Neymar") %>%
  filter(result == "Goal") %>%
  count(date, team) %>%
  ggplot() +
  geom_point(aes(date, n, color = team)) +
  labs(title = "Neymar: Barcelona vs PSG",
       color = "", x = "", y = "Liczba strzelonych\nbramek w meczu")

# skąd strzela
data %>%
  filter(player == "Neymar") %>%
  filter(result == "Goal") %>%
  ggplot() +
  geom_point(aes(x_coordinate, y_coordinate, color = team)) +
  # boisko
  xlim(c(0, 1)) +
  ylim(c(0, 1)) +
  geom_rect(aes(xmin = 0, ymin = 0, xmax = 1, ymax = 1), fill = NA, color = "black") +
  geom_rect(aes(xmin = 0, ymin = 0, xmax = 0.5, ymax = 1), fill = NA, color = "black") +
  geom_rect(aes(xmin = 0, ymin = 0.2, xmax = 0.157, ymax = 0.8), fill = NA, color = "black") +
  geom_rect(aes(xmin = 1, ymin = 0.2, xmax = 1-0.157, ymax = 0.8), fill = NA, color = "black") +
  labs(title = "Neymar: Barcelona vs PSG", x = "", y = "", color = "") +
  theme_void()

top_strikers <- data %>%
  filter(result == "Goal") %>%
  count(player, sort = T) %>%
  top_n(12, n) %>%
  pull(player)

data %>%
  filter(player %in% top_strikers) %>%
  filter(result == "Goal") %>%
  ggplot() +
  geom_point(aes(x_coordinate, y_coordinate, color = player),
             show.legend = FALSE) +
  facet_wrap(~player) +
  # boisko
  xlim(c(0, 1)) +
  ylim(c(0, 1)) +
  geom_rect(aes(xmin = 0, ymin = 0, xmax = 1, ymax = 1), fill = NA, color = "black") +
  geom_rect(aes(xmin = 0, ymin = 0, xmax = 0.5, ymax = 1), fill = NA, color = "black") +
  geom_rect(aes(xmin = 0, ymin = 0.2, xmax = 0.157, ymax = 0.8), fill = NA, color = "black") +
  geom_rect(aes(xmin = 1, ymin = 0.2, xmax = 1-0.157, ymax = 0.8), fill = NA, color = "black") +
  labs(title = "Skąd padają bramki strzelone przez...", x = "", y = "") +
  theme_void()

data %>%
  filter(player == "Lionel Messi", result == "Goal") %>%
  filter(y_coordinate == min(y_coordinate)) %>%
  t() %>%
  as.data.frame() %>% rownames_to_column() %>% set_names(c("Feature", "Value")) %>%
  kable()  %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

data %>%
  filter(player %in% top_strikers) %>%
  filter(result == "Goal") %>%
  mutate(distance = sqrt((x_coordinate-1)^2 + (y_coordinate-0.5)^2)) %>%
  group_by(player) %>%
  mutate(m_dist = mean(distance)) %>%
  ungroup() %>%
  arrange(m_dist) %>%
  mutate(player = fct_inorder(player)) %>%
  ggplot() +
  geom_boxplot(aes(player, distance, fill = player), color = "black",
               show.legend = FALSE) +
  coord_flip() +
  labs(title = "Z jakiej odległości od bramki padają gole?",
       x = "", y  = "Odległośc [skala względna]")

data %>%
  filter(result == "Goal") %>%
  mutate(distance = sqrt((x_coordinate-1)^2 + (y_coordinate-0.5)^2)) %>%
  filter(distance == max(distance)) %>%
  t() %>%
  as.data.frame() %>% rownames_to_column() %>% set_names(c("Feature", "Value")) %>%
  kable() %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

data %>%
  select(league, home_goals, away_goals, match_id) %>%
  distinct(match_id, .keep_all = TRUE) %>%
  count(league, away_goals, home_goals, sort = T) %>%
  group_by(league) %>%
  mutate(p = 100*n/sum(n)) %>%
  ungroup() %>%
  ggplot() +
  geom_tile(aes(home_goals, away_goals, fill = p), color = "black", show.legend = FALSE) +
  geom_text(aes(home_goals, away_goals, label = sprintf("%.f%%", p))) +
  scale_x_continuous(breaks = 0:10, limits = c(-1,11)) +
  scale_y_continuous(breaks = 0:10, limits = c(-1,11)) +
  scale_fill_distiller(palette = "Reds", direction = 1) +
  labs(title = "Rozkład wyników w meczach",
       x = "Gospodarze", y = "Goście") +
  facet_wrap(~league)

data_pairs <- data %>%
  select(home, away, league, home_goals, away_goals, match_id) %>%
  distinct(match_id, .keep_all = TRUE) %>%
  mutate(hG = if_else(home < away, home_goals, away_goals),
         aG = if_else(home < away, away_goals, home_goals),
         h = if_else(home < away, home, away),
         a = if_else(home < away, away, home)) %>%
  count(h, hG, a, aG) %>%
  group_by(h, a) %>%
  mutate(p = 100*n/sum(n)) %>%
  ungroup()

A_team = "Manchester City"
B_team = "Liverpool"

data_sel <- data_pairs %>%
  filter(h %in% c(A_team, B_team),
         a %in% c(A_team, B_team))

home_team <- data_sel %>% pull(h) %>% unique()
away_team <- data_sel %>% pull(a) %>% unique()

data_sel %>%
  ggplot() +
  geom_tile(aes(hG, aG, fill = p), color = "black", show.legend = FALSE) +
  geom_text(aes(hG, aG, label = sprintf("%.f%%", p))) +
  scale_x_continuous(breaks = 0:10, limits = c(-1,11)) +
  scale_y_continuous(breaks = 0:10, limits = c(-1,11)) +
  scale_fill_distiller(palette = "Reds", direction = 1) +
  labs(title = paste("Rozkład wyników w meczach\n", home_team, "-", away_team),
       x = home_team, y = away_team)

sel_team = "Barcelona"

# bilans rozgrywek
data_pairs %>%
  filter(h == sel_team | a == sel_team) %>%
  mutate(sel_goals = if_else(h == sel_team, hG, aG),
         other_goals =  if_else(h == sel_team, aG, hG),
         other =  if_else(h == sel_team, a, h)) %>%
  select(other, sel_goals, other_goals, n) %>%
  gather("key", "val", -other, -n) %>%
  mutate(key = factor(key, levels = c("sel_goals", "other_goals"),
                      labels = c("Bramki strzelone", "Bramki stracone"))) %>%
  ggplot() +
  geom_tile(aes(other, val, fill = n), color = "black") +
  scale_y_continuous(breaks = 0:10) +
  scale_fill_distiller(palette = "Reds", direction = 1) +
  facet_wrap(~key, ncol = 2) +
  theme(axis.text.x = element_text(angle = 90, vjust = 0, hjust = 1)) +
  labs(title = paste0("Liczba strzelonych/straconych bramek: ", sel_team),
       x = "Przeciwnik", y = "", fill = "Liczba bramek")

data_pairs %>%
  filter(h == sel_team | a == sel_team) %>%
  mutate(sel_goals = if_else(h == sel_team, hG, aG),
         other_goals =  if_else(h == sel_team, aG, hG),
         other =  if_else(h == sel_team, a, h)) %>%
  select(other, sel_goals, other_goals, n) %>%
  mutate(sel_goals = n * sel_goals,
         other_goals = n * other_goals) %>%
  group_by(other) %>%
  summarise(sel_goals = sum(sel_goals)/n(),
            other_goals = sum(other_goals)/n()) %>%
  ungroup() %>%
  mutate(delta = sel_goals - other_goals) %>%
  arrange(desc(delta)) %>%
  mutate(other = fct_inorder(other) %>% fct_rev()) %>%
  ggplot() +
  geom_col(aes(other, sel_goals), fill = "green") +
  geom_text(aes(other, sel_goals, label = sprintf("%.2f", sel_goals)), hjust = -0.1) +
  geom_col(aes(other, -other_goals), fill = "red") +
  geom_text(aes(other, -other_goals, label = sprintf("%.2f", other_goals)), hjust = 1.01) +
  geom_point(aes(other, delta), color = "black", size = 2) +
  # geom_text(aes(other, delta, label = sprintf("%.2f", delta)), hjust = -0.2) +
  coord_flip() +
  labs(title = paste0("Średnia liczba bramek strzelonych/straconych: ", sel_team),
       subtitle = "Zielony = bramki strzelone\nCzerwony = bramki stracone",
       x = "Przeciwnik", y = "")

passes_graph <- data %>%
  filter(result == "Goal") %>%
  filter(player %in% top_strikers) %>%
  filter(!is.na(assist_player)) %>%
  count(assist_player, player) %>%
  select(from = assist_player, to = player, weight = n) %>%
  graph_from_data_frame(directed = TRUE)


plot(passes_graph,
     vertex.color = if_else(V(passes_graph)$name %in% top_strikers, "blue", "lightblue"),
     vertex.size = if_else(V(passes_graph)$name %in% top_strikers, 7, 3),
     vertex.label.cex = 1,
     vertex.label.color =  if_else(V(passes_graph)$name %in% top_strikers, "gray80", "gray20"),
     edge.arrow.size = 0,
     edge.arrow.width = E(passes_graph)$weight,
     edge.width = E(passes_graph)$weight,
     edge.curved = TRUE)

library(networkD3)

passes_graph_d3 <- igraph_to_networkD3(passes_graph)

forceNetwork(Links = passes_graph_d3$links, Nodes = passes_graph_d3$nodes,
             Source = 'source', Target = 'target',
             NodeID = 'name', Group = 'name',
             linkWidth = passes_graph_d3$links$value,
             fontSize = 11,
             opacityNoHover = 0.6,
             zoom = TRUE)