20240201 NHL cumulative goal differential.R

# -- Created by Owen Thompson (@OTStats)
# 2024-02-01

# -- Load libraries
library(tidyverse)
library(ggpath)

# -- Import data
# Data downloaded on 2/2/24 from Moneypuck.com
df <- read_csv("Downloads/shots_2023 5.csv")


# Data transformation
shot_df <- 
  df %>% 
  transmute(
    game_id, 
    period, 
    team = homeTeamCode, 
    opp = awayTeamCode, 
    team_goals = homeTeamGoals, 
    opp_goals = awayTeamGoals, 
    time
  ) %>% 
  bind_rows(
    df %>% 
      transmute(
        game_id, 
        period, 
        opp = homeTeamCode, 
        team = awayTeamCode, 
        opp_goals = homeTeamGoals, 
        team_goals = awayTeamGoals, 
        time
      )
  )


raw_game_goals_df <-  shot_df %>% 
  group_by(game_id, team, opp) %>% 
  filter(max(time) == time) %>% 
  distinct() %>% 
  group_by(team) %>% 
  mutate(game_number = row_number(), 
         goal_diff = team_goals - opp_goals, 
         cum_goal_diff = cumsum(goal_diff))

# Set starting 
team_start_df <- 
  raw_game_goals_df %>% 
  distinct(team) %>% 
  transmute(team, 
            game_number = 0, 
            cum_goal_diff = 0)

game_goals_df <- bind_rows(team_start_df, game_goals_df) %>% 
  arrange(team, game_number)


# -- Pull in NHL logos
# Courtesy of Ivo Villanueva CSV
# Need to manually adjust some team abbrevation in order to join tables
data_git_nhl <- read.csv("https://raw.githubusercontent.com/IvoVillanueva/NHL/main/dataNHL.csv")
logos_df <- 
  data_git_nhl %>% 
  transmute(
    team = str_to_upper(url_abr), 
    logo = espn_logo, 
    primary
  ) %>% 
  distinct() %>% 
  add_row(team = "SEA") %>% 
  mutate(
    team = 
      case_when(
        team == "TB" ~ "TBL", 
        team == "VGS" ~ "VGK", 
        team == "LA" ~ "LAK", 
        team == "NJ" ~ "NJD", 
        team == "SJ" ~ "SJS", 
        team == "SEA" ~ "SEA", 
        TRUE ~ team
        )
  ) %>% 
  mutate(logo = case_when(
    team == "VGK" ~ "https://a.espncdn.com/i/teamlogos/nhl/500/vgs.png", 
    team == "SEA" ~ "https://a.espncdn.com/i/teamlogos/nhl/500/sea.png", 
    TRUE ~ logo))

# # Validate logos for each team
# shot_df %>% 
#   distinct(team) %>% 
#   left_join(logos_df)

# Set logo URLs to most recent game number
plot_image_df <- 
  game_goals_df %>% 
  group_by(team) %>% 
  filter(max(game_number) == game_number) %>% 
  left_join(logos_df)

# -- Plot image
game_goals_df %>% 
  left_join(logos_df) %>% 
  ggplot(aes(x = game_number, y = cum_goal_diff, group = team)) + 
  geom_hline(yintercept = 0) + 
  geom_line(aes(color = primary), linewidth = 1) + 
  scale_color_identity() + 
  geom_from_path(
    data = plot_image_df, 
    aes(x = game_number, y = cum_goal_diff, path = logo), width = 0.03
  ) + 
  labs(
    title = "Cumulative Goal Differential by Team", 
    subtitle = "2024 Regular Season | Data as of 2/1/24", 
    x = "Game Number", 
    y = "Cumulative Goal Differential", 
    caption = "Viz: @OTStats | Data: Moneypuck.com"
  ) + 
  scale_x_continuous(minor_breaks = NULL) + 
  theme_minimal() + 
  theme(
    plot.title = element_text(hjust = 0.5, size = 16, face = "bold"), 
    plot.subtitle = element_text(hjust = 0.5, size = 12), 
    axis.title = element_text(face = "bold")
  )

ggplot2::ggsave(filename = "20240201 Cumulative Goal Differential by NHL Team.png", width = 5, height = 3, dpi = "retina")


# # -- Alternative viz 1: use {gghighlight} to highlight specific teams or highlight by facet
# library(gghighlight)
# game_goals_df %>% 
#   ggplot(aes(x = game_number, y = cum_goal_diff, group = team)) + 
#   geom_line() + 
#   geom_from_path(
#     data = plot_image_df, 
#     aes(x = game_number, y = cum_goal_diff, path = logo), width = 0.03
#   ) + 
#   gghighlight() + 
#   facet_wrap(~team)


# # -- Alternative viz 2: {gghighlight} to spotlight particular teams
# game_goals_df %>% 
#   left_join(logos_df) %>% 
#   ggplot(aes(x = game_number, y = cum_goal_diff, group = team)) + 
#   geom_line(aes(color = primary)) + 
#   geom_from_path(
#     data = plot_image_df, 
#     aes(x = game_number, y = cum_goal_diff, path = logo), width = 0.03
#   ) + 
#   gghighlight(max(cum_goal_diff) > 25) + 
#   scale_color_identity()