In systats/lolR: Data Sourcing

Packages

pacman::p_load(dplyr, ggplot2, rvest, purrr, xml2, stringr, anytime, magrittr, tidyr, tibble, devtools)
devtools::load_all()

Load Data

load("data/matches_acs.Rdata")
load("data/matches.Rdata")

Reorganize Gamepedia data

glimpse(matches)

matches_gp <- matches %>% 
  dplyr::select(
    gp_date = date,
    gp_patch = patch, 
    gp_patch_link = patch_link,
    gp_win = win,
    blue_team, blue_team_link, red_team, red_team_link,
    players, 
    blue_gold,
    red_gold,
    contains("tourn"),
    game_id, game_hash, match_url, api_url, vod
  ) %>% 
  mutate(gp_patch = ifelse(gp_patch == "-", NA, gp_patch)) %>% 
  mutate(game_id = as.integer(game_id))

matches_gp %>% glimpse

Merge matches_gp and matches_acs

lol_matches <- matches_gp %>% 
  left_join(matches_acs)

Explore Data

How many with game id?

lol_matches %>% 
  count(is.na(game_id))

lol_matches_complete <- lol_matches %>% 
  filter(!is.na(game_id) & !is.na(player_stats))


join_player <- function(x){
  x$players[[1]] %>% 
    ### naming inconsistency!!!
    #rename(red_players = red_player, blue_players = blue_player) %>%
    #right_join(x$players[[1]])
    bind_cols(x$player_stats[[1]])
}

join_player_safely <- purrr::safely(join_player)

joined_player <- lol_matches_complete %>% 
  split(1:nrow(.)) %>%
  map(join_player_safely)

# joined_player %>% 
#   map("error")

lol_matches_complete$player_final <- joined_player %>% 
  map("result")

lol_matches_complete <- lol_matches_complete %>% 
  select(-players, -player_stats)

save(lol_matches_complete, file = "data/lol_matches_complete.RData")
glimpse(lol_matches_complete)

lol_matches_complete$players[[1]]
lol_matches_complete$player_stats[[1]]
lol_matches_complete$player_final[[1]]

# nn <- lol_matches_complete %>% 
#   split(1:nrow(.)) %>%
#   map_int({~ .x$player_final[[1]] %>% 
#       .$red_champion_id %>% is.na() %>% 
#       ifelse(is.null(.), NA, .) %>%
#       table()[1]
#   })
# 
# 
# .x$player_final[[1]] %>% 
#       .$red_champion_id %>% is.na() %>% 
#       table()[1]

lol_matches_complete$player_final[[1]] %>% 
  select(contains("player")) %>% 
  glimpse()