2023-01-17 21:56:21 +01:00
|
|
|
import pandas as pd
|
|
|
|
|
from FuzzySystemFunctions import *
|
|
|
|
|
from numpy import inf, nan, nan_to_num
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
names_to_shortcuts = {
|
2023-01-31 10:54:42 +01:00
|
|
|
"Evil Geniuses": "EG",
|
2023-01-17 21:56:21 +01:00
|
|
|
"TSM": "TSM",
|
|
|
|
|
"Cloud9": "C9",
|
|
|
|
|
"Immortals Progressive": "IMT",
|
|
|
|
|
"100 Thieves": "100",
|
|
|
|
|
"Dignitas": "DIG",
|
|
|
|
|
"CLG": "CLG",
|
|
|
|
|
"FlyQuest": "FLY",
|
|
|
|
|
"Team Liquid Honda": "TL",
|
|
|
|
|
"Golden Guardians": "GG",
|
|
|
|
|
"EXCEL": "XL",
|
|
|
|
|
"G2 Esports": "G2",
|
|
|
|
|
"Fnatic": "FNC",
|
|
|
|
|
"Astralis": "AST",
|
|
|
|
|
"Schalke 04 Esports": "S04",
|
|
|
|
|
"SK Gaming": "SK",
|
|
|
|
|
"Rogue": "RGE",
|
|
|
|
|
"MAD Lions": "MAD",
|
|
|
|
|
"Misfits Gaming": "MSF",
|
|
|
|
|
"Team Vitality": "VIT",
|
|
|
|
|
"Maturalni Forsaken": "FSK",
|
|
|
|
|
"AGO Rogue": "RGO",
|
|
|
|
|
"Illuminar Gaming": "IHG",
|
|
|
|
|
"Devils.One": "DV1",
|
|
|
|
|
"Komil&Friends": "KNF",
|
|
|
|
|
"Zero Tenacity": "Z10",
|
|
|
|
|
"Gentlemen's Gaming": "GGM",
|
|
|
|
|
"Goskilla": "GSK",
|
|
|
|
|
"Topo Centras Iron Wolves": "WOLF",
|
|
|
|
|
"Team ESCA Gaming": "ESCA"
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def get_win_team_name(row):
|
|
|
|
|
if row["blueTeam_result"] == "loss":
|
|
|
|
|
return row["redTeam_name"]
|
|
|
|
|
else:
|
|
|
|
|
return row["blueTeam_name"]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def get_win_ratio(df, team_name, number_of_games):
|
|
|
|
|
df = df[(df["blueTeam_name"] == team_name) | (df["redTeam_name"] == team_name)]
|
|
|
|
|
df = df.head(number_of_games)
|
2023-01-31 10:54:42 +01:00
|
|
|
|
2023-01-17 21:56:21 +01:00
|
|
|
return len(df[df["winTeam_name"] == team_name]) / len(df)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def get_last_5_games_win_ratio(df, team_name, number_of_games):
|
|
|
|
|
df = df[(df["blueTeam_name"] == team_name) | (df["redTeam_name"] == team_name)]
|
|
|
|
|
df = df.head(number_of_games)
|
|
|
|
|
df = df.tail(5)
|
|
|
|
|
|
|
|
|
|
return len(df[df["winTeam_name"] == team_name]) / len(df)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def get_games_ids(df, team_name, number_of_games):
|
|
|
|
|
df = df[(df["blueTeam_name"] == team_name) | (df["redTeam_name"] == team_name)]
|
|
|
|
|
df = df.head(number_of_games)
|
2023-01-31 10:54:42 +01:00
|
|
|
|
2023-01-17 21:56:21 +01:00
|
|
|
return df["id"]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def get_kdas(df_player_stats, games_ids, team_name):
|
|
|
|
|
team_name_shortcut = names_to_shortcuts[team_name]
|
|
|
|
|
kda = []
|
|
|
|
|
|
|
|
|
|
for game_id in games_ids:
|
|
|
|
|
game_info = df_player_stats[df_player_stats["game_id"] == game_id]
|
|
|
|
|
game_info = game_info[game_info["summonerName"].str.startswith(team_name_shortcut)]
|
|
|
|
|
|
|
|
|
|
players_kdas = (game_info["kills"] + game_info["assists"]) / game_info["deaths"]
|
|
|
|
|
players_kdas = players_kdas.values
|
|
|
|
|
players_kdas[players_kdas == inf] = 10
|
|
|
|
|
players_kdas = nan_to_num(players_kdas, nan=1)
|
|
|
|
|
players_kdas = players_kdas.mean()
|
|
|
|
|
|
|
|
|
|
kda.append(players_kdas)
|
|
|
|
|
|
|
|
|
|
return sum(kda) / len(kda)
|
|
|
|
|
|
|
|
|
|
|
2023-01-31 10:54:42 +01:00
|
|
|
def get_prediction(team_A, team_B, number_of_games_A, number_of_games_B, league, fuzzy_system, team_a_win, verbose):
|
2023-01-17 21:56:21 +01:00
|
|
|
df_lol = pd.read_csv(f"matchHistory.csv")
|
|
|
|
|
df_lol = df_lol[df_lol["tournament"] == league]
|
|
|
|
|
|
|
|
|
|
df_lol["gameStarted"] = pd.to_datetime(df_lol["gameStarted"])
|
|
|
|
|
df_lol.sort_values(by="gameStarted", inplace=True)
|
|
|
|
|
df_lol["winTeam_name"] = df_lol.apply(get_win_team_name, axis=1)
|
|
|
|
|
|
|
|
|
|
df_player_stats = pd.read_csv(f"statsGame.csv")
|
|
|
|
|
df_player_stats = df_player_stats[df_player_stats["tournament"] == league]
|
|
|
|
|
team_A_games_ids = get_games_ids(df_lol, team_A, number_of_games_A)
|
|
|
|
|
team_B_games_ids = get_games_ids(df_lol, team_B, number_of_games_B)
|
|
|
|
|
|
|
|
|
|
kda_team_A = get_kdas(df_player_stats, team_A_games_ids, team_A)
|
|
|
|
|
kda_team_B = get_kdas(df_player_stats, team_B_games_ids, team_B)
|
|
|
|
|
kda_proportion = kda_team_A / kda_team_B
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
team_A_standings = get_win_ratio(df_lol, team_A, number_of_games_A) * 100
|
|
|
|
|
team_B_standings = get_win_ratio(df_lol, team_B, number_of_games_B) * 100
|
|
|
|
|
|
|
|
|
|
team_A_form = get_last_5_games_win_ratio(df_lol, team_A, number_of_games_A) * 100
|
|
|
|
|
team_B_form = get_last_5_games_win_ratio(df_lol, team_B, number_of_games_B) * 100
|
|
|
|
|
|
|
|
|
|
adv_standings = (team_A_standings - team_B_standings) / 2 + 50
|
|
|
|
|
adv_form = (team_A_form - team_B_form) / 2 + 50
|
|
|
|
|
adv_performance = (1 - (1 / (1 + kda_proportion))) * 100
|
|
|
|
|
|
2023-01-31 10:54:42 +01:00
|
|
|
if verbose is True:
|
|
|
|
|
print("\n########## PREDICTION ##########")
|
|
|
|
|
print(f"Advantage standings: {adv_standings}")
|
|
|
|
|
print(f"Advantage form: {adv_form}")
|
|
|
|
|
print(f"Advantage performance: {adv_performance}")
|
|
|
|
|
|
|
|
|
|
res = make_inference(fuzzy_system, adv_standings, adv_form, adv_performance)["RESULT"]
|
|
|
|
|
return res
|
