predykcja_analiza_sportowa/predict.ipynb

import pandas as pd

!unzip matches.zip

Archive:  matches.zip
   creating: matches/
  inflating: matches/international_matches.csv  

Projekt do przewidywania, wygranego zespołu w meczu na podstawie fifa points

Używamy datasetu:

1. Wyniki meczów podczas fifa world cup (1993-2022)

# points = pd.read_csv('ranking/fifa_ranking-2022-10-06.csv')
matches = pd.read_csv('matches/international_matches.csv')

# points = points[["country_full", "total_points", "previous_points", "rank_date"]]
matches = matches[["date", "home_team", "away_team", "home_team_fifa_rank","away_team_fifa_rank", "home_team_score", "away_team_score", "home_team_result"]]

# points

matches

	date	home_team	away_team	home_team_fifa_rank	away_team_fifa_rank	home_team_score	away_team_score	home_team_result
0	1993-08-08	Bolivia	Uruguay	59	22	3	1	Win
1	1993-08-08	Brazil	Mexico	8	14	1	1	Draw
2	1993-08-08	Ecuador	Venezuela	35	94	5	0	Win
3	1993-08-08	Guinea	Sierra Leone	65	86	1	0	Win
4	1993-08-08	Paraguay	Argentina	67	5	1	3	Lose
...	...	...	...	...	...	...	...	...
23916	2022-06-14	Moldova	Andorra	180	153	2	1	Win
23917	2022-06-14	Liechtenstein	Latvia	192	135	0	2	Lose
23918	2022-06-14	Chile	Ghana	28	60	0	0	Lose
23919	2022-06-14	Japan	Tunisia	23	35	0	3	Lose
23920	2022-06-14	Korea Republic	Egypt	29	32	4	1	Win

23921 rows × 8 columns

Na początku zrobimy naiwne założenia, zespół który ma więcej fifa points wygrywa, jeśli mają tyle samo zakładamy remis

p_true = 0
p_false = 0
for i, m in matches.iterrows():
    if m["home_team_fifa_rank"] > m["away_team_fifa_rank"] and m["home_team_result"] == "Win":
        p_true +=1
    elif m["home_team_fifa_rank"] < m["away_team_fifa_rank"] and m["home_team_result"] == "Lose":
        p_true +=1
    elif m["home_team_fifa_rank"] == m["away_team_fifa_rank"] and m["home_team_result"] == "Draw":
        p_true +=1
    else:
        p_false +=1
print("Accuracy: ", p_true/(p_true + p_false))

Accuracy:  0.21809288909326532

Teraz za pomocą regresji logistycznej i fifa points będziemy przewidywać wygrany zespół

matches = matches[["home_team_fifa_rank", "away_team_fifa_rank", "home_team_result"]]

matches["home_team_result"] = matches["home_team_result"].apply(lambda x: 2 if x == 'Win' else x)

/tmp/ipykernel_26917/2591599491.py:1: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  matches["home_team_result"] = matches["home_team_result"].apply(lambda x: 2 if x == 'Win' else x)

matches["home_team_result"] = matches["home_team_result"].apply(lambda x: 1 if x == 'Draw' else x)

/tmp/ipykernel_26917/1297296633.py:1: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  matches["home_team_result"] = matches["home_team_result"].apply(lambda x: 1 if x == 'Draw' else x)

matches["home_team_result"] = matches["home_team_result"].apply(lambda x: 0 if x == 'Lose' else x)

/tmp/ipykernel_26917/2245446894.py:1: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  matches["home_team_result"] = matches["home_team_result"].apply(lambda x: 0 if x == 'Lose' else x)

matches

	home_team_fifa_rank	away_team_fifa_rank	home_team_result
0	59	22	2
1	8	14	1
2	35	94	2
3	65	86	2
4	67	5	0
...	...	...	...
23916	180	153	2
23917	192	135	0
23918	28	60	0
23919	23	35	0
23920	29	32	2

23921 rows × 3 columns

import numpy as np
from sklearn.model_selection import train_test_split

X = matches[["home_team_fifa_rank", "away_team_fifa_rank"]]
Y = matches["home_team_result"]
data = np.array(matches)

X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.33, random_state=42)

from sklearn.datasets import load_iris
from sklearn.linear_model import LogisticRegression

clf = LogisticRegression(random_state=1).fit(X_train, y_train)
clf.score(X_train, y_train)

0.5674174829974418

y_pred = clf.predict(X_test)

from sklearn.metrics import accuracy_score
accuracy_score(y_test, y_pred)

0.5744869521155308

Dzięki zastostosowaniu regresji logistycznej uzyskaliśmy accuracy 57%, biorąc pod uwagę, że przewidywaliśmy 3 klasy wygraną, remis i przegraną nie jest to najgorszy wynik

Ranking fifa points wydaję się być dobry baselinem do modelów predykcyjnych