我正在尝试在Spyder Python 3.7上运行以下Python脚本。我对python非常陌生，这是我第一次尝试运行如此复杂的程序：
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import numpy as np
import statsmodels.api as sm
import statsmodels.formula.api as smf
import itertools
matches = pd.read_csv( '/Input/international-football-results-from-1872-to-2017/results 3.csv')
matches.head(2)
matches = matches.astype({'date':'datetime64[ns]'})
print = ( "Then I create a dataframe with some statistics by team like the sum, count and mean of score for each team.")
# create two dataframe for the home and away teams
home = matches[['home_team', 'home_score']].rename(columns={'home_team':'team', 'home_score':'score'})
away = matches[['away_team', 'away_score']].rename(columns={'away_team':'team', 'away_score':'score'})
# merge it into one and removes any corrilation between away and home teams
team_score = home.append(away).reset_index(drop=True)
team_score.head(5)
# make an aggregation of the the score column group by the team
country_info = team_score.groupby('team')['score'].agg(['sum','count','mean']).reset_index()
country_info = country_info.rename(columns={'sum':'numb_goals', 'count':'numb_matches', 'mean':'goal_avg'})
del home, away
means = matches[['home_score','away_score']].mean()
means
def weight_from_tournament(tournament):
if 'Cup' in tournament or 'Euro' in tournament or 'Copa del Rey' in tournament or 'UEFA' in tournament or 'Copa Libertadores' in tournament or 'Copa America' in tournament:
return 1;
else :
return 100;
# Weight column based on type of tournament
matches.loc[:,'weight'] = matches['tournament'].apply(weight_from_tournament)
# Mpdify weight column based on the date
matches.loc[:,'weight'] = 1 / ((2019 - matches['date'].dt.year.astype('int64'))*matches['weight'])
# Create model data
matches_model_data = pd.concat([matches[['home_team','away_team','home_score','weight']].rename(
columns={'home_team':'team', 'away_team':'opponent','home_score':'goals'}),
matches[['away_team','home_team','away_score','weight']].rename(
columns={'away_team':'team', 'home_team':'opponent','away_score':'goals'})])
#So here using the statsmodels librairies I create my model with a Poisson regression.
poisson_model = smf.glm(formula="goals ~ team + opponent", data=matches_model_data,
family=sm.families.Poisson(), freq_weights=matches_model_data['weight'].values).fit()
print = ("Now it's time to make a function that can return the result of a game, there is some information that we need to know : ")
def get_proba_match(foot_model, team1, team2, max_goals=10):
# Get the average goal for each team
t1_goals_avg = foot_model.predict(pd.DataFrame.apply(data={'team': team1, 'opponent': team2}, index=[1])).values[0]
t2_goals_avg = foot_model.predict(pd.DataFrame.apply(data={'team': team2, 'opponent': team1}, index=[1])).values[0]
# Get probability of all possible score for each team
team_pred = [[poisson.pmf(i, team_avg) for i in range(0, max_goals+1)] for team_avg in [t1_goals_avg, t2_goals_avg]]
# Do the product of the 2 vectors to get the matrix of the match
match = np.outer(np.array(team_pred[0]), np.array(team_pred[1]))
# Get the proba for each possible outcome
t1_wins = np.sum(np.tril(match, -1))
draw = np.sum(np.diag(match))
t2_wins = np.sum(np.triu(match, 1))
result_proba = [t1_wins, draw, t2_wins]
# Adjust the proba to sum to one
result_proba = np.array(result_proba)/ np.array(result_proba).sum(axis=0,keepdims=1)
team_pred[0] = np.array(team_pred[0])/np.array(team_pred[0]).sum(axis=0,keepdims=1)
team_pred[1] = np.array(team_pred[1])/np.array(team_pred[1]).sum(axis=0,keepdims=1)
return result_proba, [np.array(team_pred[0]), np.array(team_pred[1])]
def get_match_result(foot_model, team1, team2, elimination=False, max_draw=50, max_goals=10):
# Get the proba
proba, score_proba = get_proba_match(foot_model, team1, team2, max_goals)
# Get the result, if it's an elimination game we have to be sure the result is not draw
results = pd.Series([np.random.choice([team1, 'draw', team2], p=proba) for i in range(0,max_draw)]).value_counts()
result = results.index[0] if not elimination or (elimination and results.index[0] != 'draw') else results.index[1]
# If the result is not a draw game then we calculate the score of the winner from 1 to the max_goals
# and the score of the looser from 0 to the score of the winner
if (result != 'draw'):
i_win, i_loose = (0,1) if result == team1 else (1,0)
score_proba[i_win] = score_proba[i_win][1:]/score_proba[i_win][1:].sum(axis=0,keepdims=1)
winner_score = pd.Series([np.random.choice(range(1, max_goals+1), p=score_proba[i_win]) for i in range(0,max_draw)]).value_counts().index[0]
score_proba[i_loose] = score_proba[i_loose][:winner_score]/score_proba[i_loose][:winner_score].sum(axis=0,keepdims=1)
looser_score = pd.Series([np.random.choice(range(0, winner_score), p=score_proba[i_loose]) for i in range(0,max_draw)]).value_counts().index[0]
score = [winner_score, looser_score]
# If it's a draw then we calculate a score and repeat it twice
else:
score = np.repeat(pd.Series([np.random.choice(range(0, max_goals+1), p=score_proba[0]) for i in range(0,max_draw)]).value_counts().index[0],2)
looser = team2 if result == team1 else team1 if result != 'draw' else 'draw'
return result, looser, score
print = ("Now let's try it !")
print(get_match_result(poisson_model, 'Gabon', 'Togo'))
print(get_match_result(poisson_model, 'France', 'Togo', elimination=True))
print(get_match_result(poisson_model, 'Argentina', 'Germany'))
print(get_match_result(poisson_model, 'Brazil', 'Vatican', max_goals=20))
print(get_match_result(poisson_model, 'England', 'Morocco'))
print(get_match_result(poisson_model, 'Iran', 'Japan'))
但是，出现以下错误：
print(get_match_result(poisson_model, 'Gabon', 'Togo'))
Traceback (most recent call last):
File "<ipython-input-72-3287e60bece4>", line 1, in <module>
print(get_match_result(poisson_model, 'Gabon', 'Togo'))
TypeError: 'DataFrame' object is not callable
我已从https://www.kaggle.com/nathanlauga/which-team-will-win-the-2018-fifa-world-cup
修改了脚本如果用叉子分叉笔记本电脑并使用Kaggle在线控制台，运行会很好。
任何和所有建议将不胜感激。
答案 0 :(得分：1)
您的问题仅在语法上。
在Python中，您不想表达print = ('something')，因为这意味着您正在为名为print的变量分配值。相反，您必须做print('something')
我希望它能帮助