Geoguessr game analysis¶

from os import environ
import atoti as tt

config = "mybinder-config.yml" if "JUPYTERHUB_SERVICE_PREFIX" in environ else "local-config.yml"
session = tt.create_session("Geoguessr", config=config)
print(session.url)

scores = session.read_csv("data/scores.csv", keys=["GameId", "Round", "Player"])
games = session.read_csv("data/games.csv",  keys=["GameId"])
rounds = session.read_csv("data/rounds.csv", keys=["GameId", "Round"])
continents = session.read_csv("data/continents.csv", keys=["Country"], sep=",")
scores.join(games, mapping={"GameId": "GameId"})
scores.join(rounds, mapping={"GameId": "GameId", "Round": "Round"})
rounds.join(continents, mapping={"Country": "Country"})

session.stores.schema

cube = session.create_cube(scores)

m, lvl, h = cube.measures, cube.levels, cube.hierarchies

h["Games"] = [scores["GameId"], scores["Round"]]
h["Round Number"] = { "Round Number":  scores["Round"]}
del h["GameId"]
del h["Round"]

h["Round Number"] = { "Round Number":  scores["Round"]}

h["Geography"] = [continents["Continent"], rounds["Country"]]
del h["Continent"]
del h["Country"]
h["Geography"].dimension = "Geography"

h

m["score"] = tt.agg.mean(
    tt.agg.sum(scores["Score"]),
    scope=tt.scope.origin(lvl["Player"], lvl["GameId"])
)
m["score"].formatter = "DOUBLE[#,###]"

m["avg score"] = tt.agg.mean(scores["Score"])

session.visualize()

session.visualize("Average Score per country")

session.visualize("Average Score per continent")

m["one"] = 1
m["Number of rounds"] = tt.agg.sum(m["one"], scope=tt.scope.origin(lvl["Round"]))

session.visualize("Most seen countries")

session.visualize("Most seen continent")

m["find country rate"] = tt.filter(m["contributors.COUNT"], lvl["HasCountry"] == True) / m["contributors.COUNT"]
m["find country rate"].formatter = "DOUBLE[0.00%]"

session.visualize("Find country rate per user")

m["Best score"] = tt.agg.max(m["score"], scope=tt.scope.origin(lvl["GameId"], lvl["Player"]))
m["Worst score"] = tt.agg.min(m["score"], scope=tt.scope.origin(lvl["GameId"], lvl["Player"]))

cube.query(m["Best score"], levels=lvl["Player"])

m["Winner"] = tt.where(
    lvl["GameId"] != None,
    tt.agg.max_member(m["score"], lvl["Player"])
)