Jeopardy Survival Plot

library(tidyverse)
library(rvest)
library(survival)
library(GGally)
library(stringr)
library(survminer)

game_data = tibble(
  season = 1:33
) %>%
  mutate(link = paste0("http://j-archive.com/showseason.php?season=", season),
         html = map(link, read_html),
         table = map(html, html_node, "table"),
         dat = map(table, html_table)) %>%
  select(season, dat) %>%
  unnest()

results <- game_data %>%
  transmute(airdate = X1,
            contestants = X2,
            extra = str_to_lower(X3)) %>%
  filter(!str_detect(extra, "championship"), !str_detect(extra, "tournament")) %>%
  separate(contestants, paste0("contestant", 1:3), "vs\\.") %>%
  group_by(contestant1) %>% 
  summarize(wins = n())
  
surv <- Surv(time = results$wins)
fit <- survfit(surv~1, data = results)

ggsurvplot(fit, data = results)[[1]] +
  geom_point() +
  xlab("Wins") +
  labs(title = "Jeopardy! Survival Curve",
       subtitle = paste0(nrow(results), " games"),
       caption = "Data: j-archive.com\nherbsusmann.com") +
  scale_y_continuous(breaks = seq(0, 1, 0.1)) +
  theme(plot.caption = element_text(size = 10),
        legend.position = "none")