tennis

tennis

rm(list = ls())

library(tidyverse)
library(lubridate)


players <- read_csv("https://raw.githubusercontent.com/JeffSackmann/tennis_atp/master/atp_players.csv")

decades <- c("70", "80", "90", "00", "10", "20")

paths <- paste0("https://raw.githubusercontent.com/JeffSackmann/tennis_atp/master/atp_rankings_",
                decades,
                "s.csv") %>% 
  as.list()

rm(decades)

paths


atp <- map_dfr(.x = paths,
                .f = read_csv)

rm(paths)


atp <- atp %>% 
  mutate(date = parse_date_time(as.character(ranking_date),
                                orders = "ymd"))
atp


atp <- atp %>% 
  tidylog::left_join(.,
                     players,
                     by = c("player" = "player_id"))

rm(players)


atp <- atp %>% 
  mutate(name_full = paste(name_first, name_last))



min_rank <- min(atp$rank)
max_rank <- max(atp$rank)



atp %>% 
  filter(name_full == "Novak Djokovic") %>% 
  ggplot(.,
         aes(x = date,
             y = rank)) +
  geom_line() +
  ylim(max_rank, min_rank)


atp <- atp %>% 
  group_by(player) %>% 
  mutate(min_rank = min(rank),
         max_rank = max(rank)) %>% 
  ungroup()

atp %>% 
  filter(min_rank == 10) %>% 
  ggplot(.,
         aes(x = date,
             y = rank)) +
  geom_line(aes(color = factor(name_full))) +
  ylim(100, min_rank)


atp %>% 
  filter(date >= "2000-01-01") %>% 
  filter(name_full %in% c("Novak Djokovic",
                          "Rafael Nadal",
                          "Roger Federer")) %>% 
  ggplot(.,
         aes(x = date,
             y = rank)) +
  geom_line(aes(color = name_full)) +
  ylim(max_rank, min_rank)


big_three <- atp %>% 
  filter(date >= "2000-01-01") %>% 
  filter(name_full %in% c("Novak Djokovic",
                          "Rafael Nadal",
                          "Roger Federer"))

others <- atp %>% 
  filter(date >= "2000-01-01") %>% 
  filter(!name_full %in% c("Novak Djokovic",
                           "Rafael Nadal",
                           "Roger Federer")) %>% 
  filter(min_rank <= 100)

min_rank <- 1
max_rank <- max(others$rank)


ggplot() +
  geom_line(data = others, aes(x = date,
                               y = rank,
                               group = player),
            size = .001, color = "grey90") +
  ylim(max_rank, min_rank) +
  theme_bw()



top10 <- atp %>% 
  filter(rank < 11) %>% 
  filter(date >= "2000-01-01")
min_rank <- 1
max_rank <- max(top10$rank)


ggplot() +
  geom_line(data = top10, aes(x = date,
                              y = rank,
                              group = player),
            size = .001, color = "grey90") +
  ylim(max_rank, min_rank) +
  theme_bw()


top10 <- top10 %>% 
  arrange(player, date) %>% 
  group_by(player) %>% 
  mutate(rank_roll = zoo::rollmean(rank,
                                   k = 5,
                                   fill = NA,
                                   align = "right")) %>% 
  ungroup()

top10 %>% 
  filter(date >= "2000-01-01") %>% 
  ggplot() +
  geom_line(data = top10, aes(x = date,
                              y = rank_roll,
                              group = player),
            size = .001, color = "grey90") +
  ylim(max_rank, min_rank) +
  theme_bw()