0-tidy.R

library(tidyjson)
library(tidyverse)
library(magrittr)


json_df <- read_json("data/raw/db.json", "jsonl")

# ------------------------------------------------------------------------------
# Separate data into two "tidy" tables, each with there own observational unit

players <- 
  json_df %>%
  spread_values(
    id         = json_chr("uniqueId"),
    name       = json_chr("displayName"),
    platform   = json_chr("platform", "name"),
  # start_date = json_dbl("createdAt"),  # Not sure what this means
    shots      = json_dbl("stats", "shots"),
    saves      = json_dbl("stats", "saves"),
    mvps       = json_dbl("stats", "mvps"),
    goals      = json_dbl("stats", "goals"),
    assists    = json_dbl("stats", "assists"),
    wins       = json_dbl("stats", "wins")
  )

# anytime::anytime(1468013823)  # Convert createdAt if we decide to use it

# This takes a few minutes
rank <- 
  json_df %>%
  spread_values(
    id = json_chr("uniqueId")
  ) %>%
  enter_object("rankedSeasons") %>%
  spread_all()


# ------------------------------------------------------------------------------
# Reshape the rank table so that it follows "tidy" methodology

rank %<>%
  gather(key, val, -c(document.id, id)) %>%
  separate(key, 
           c("season", "game_type", "stats"), 
           sep = "\\.") %>%
  spread(stats, val) %>%
  rename(
    matches_played = matchesPlayed, 
    mmr            = rankPoints
  )


# ------------------------------------------------------------------------------
# Remove duplicates and NAs to make the data easier to analyze

players %<>%
  drop_na() %>%
  distinct(id, .keep_all = TRUE)

rank %<>%
  drop_na() %>%
  distinct()


saveRDS(players, "data/players.Rds")
saveRDS(rank, "data/rank.Rds")