python案例如何预测点球大战胜负走向？

wen python案例 2026-10-11 3

Python案例：预测点球大战胜负走向

点球大战预测是体育数据分析中一个有趣的话题,下面我用一个完整的Python案例来演示如何基于历史数据预测点球大战的胜负。

核心思路

点球大战胜负受以下因素影响：

因素	说明
球员罚球能力	历史命中率
门将扑救能力	历史扑救率
心理压力	关键轮次（第5轮）命中率下降
先罚优势	先罚球队胜率约60%
球队整体实力	FIFA排名等

完整代码实现

数据准备

import numpy as np
import pandas as pd
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score, classification_report
import matplotlib.pyplot as plt
# 模拟历史点球大战数据（真实场景可替换为实际数据）
np.random.seed(42)
def generate_match_data(n=2000):
    data = []
    for _ in range(n):
        # 球队A、B的球员平均命中率
        a_accuracy = np.random.uniform(0.6, 0.9)
        b_accuracy = np.random.uniform(0.6, 0.9)
        # 门将扑救率
        a_keeper = np.random.uniform(0.15, 0.35)
        b_keeper = np.random.uniform(0.15, 0.35)
        # 先罚球队 (1=A先罚, 0=B先罚)
        first = np.random.choice([0, 1])
        # 大赛经验值 (0-10)
        a_exp = np.random.randint(0, 11)
        b_exp = np.random.randint(0, 11)
        # 计算每轮命中概率（考虑压力：第5轮×0.9）
        def hit_prob(acc, keeper):
            base = acc * (1 - keeper)
            return base
        # 模拟1000次点球大战计算胜率
        wins = 0
        for _ in range(200):
            score_a, score_b = 0, 0
            for r in range(5):
                pressure = 0.9 if r == 4 else 1.0
                # A罚球
                if np.random.rand() < hit_prob(a_accuracy, b_keeper) * pressure:
                    score_a += 1
                # B罚球
                if np.random.rand() < hit_prob(b_accuracy, a_keeper) * pressure:
                    score_b += 1
            # 平局进入突然死亡
            while score_a == score_b:
                if np.random.rand() < hit_prob(a_accuracy, b_keeper) * 0.85:
                    score_a += 1
                if np.random.rand() < hit_prob(b_accuracy, a_keeper) * 0.85:
                    score_b += 1
            if score_a > score_b:
                wins += 1
        win_rate_a = wins / 200
        label = 1 if win_rate_a > 0.5 else 0  # 1=A胜
        data.append([a_accuracy, b_accuracy, a_keeper, b_keeper,
                     first, a_exp, b_exp, label])
    return pd.DataFrame(data, columns=[
        'a_accuracy', 'b_accuracy', 'a_keeper', 'b_keeper',
        'a_first', 'a_exp', 'b_exp', 'label'])
df = generate_match_data(2000)
print(df.head())
print(f"数据分布：\n{df['label'].value_counts()}")

特征工程

# 构造更有意义的组合特征
df['accuracy_diff'] = df['a_accuracy'] - df['b_accuracy']        # 罚球能力差
df['keeper_diff'] = df['b_keeper'] - df['a_keeper']              # A门将比B门将强多少
df['exp_diff'] = df['a_exp'] - df['b_exp']                        # 经验差
# 综合实力分
df['a_power'] = df['a_accuracy'] * (1 - df['b_keeper']) + df['a_exp'] * 0.01
df['b_power'] = df['b_accuracy'] * (1 - df['a_keeper']) + df['b_exp'] * 0.01
X = df.drop('label', axis=1)
y = df['label']
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42, stratify=y)

训练模型

model = RandomForestClassifier(
    n_estimators=200,
    max_depth=8,
    random_state=42
)
model.fit(X_train, y_train)
y_pred = model.predict(X_test)
print(f"准确率: {accuracy_score(y_test, y_pred):.3f}")
print(classification_report(y_test, y_pred))
# 特征重要性
importances = pd.Series(model.feature_importances_, index=X.columns)
importances.sort_values().plot(kind='barh', figsize=(8, 6))'特征重要性')
plt.tight_layout()
plt.show()

预测具体比赛

def predict_penalty_shootout(team_a, team_b, model=model):
    """
    team_a/team_b: dict，包含 accuracy, keeper, exp, is_first
    """
    row = {
        'a_accuracy': team_a['accuracy'],
        'b_accuracy': team_b['accuracy'],
        'a_keeper': team_a['keeper'],
        'b_keeper': team_b['keeper'],
        'a_first': 1 if team_a.get('is_first', True) else 0,
        'a_exp': team_a['exp'],
        'b_exp': team_b['exp'],
    }
    # 添加工程特征
    row['accuracy_diff'] = row['a_accuracy'] - row['b_accuracy']
    row['keeper_diff'] = row['b_keeper'] - row['a_keeper']
    row['exp_diff'] = row['a_exp'] - row['b_exp']
    row['a_power'] = row['a_accuracy'] * (1 - row['b_keeper']) + row['a_exp'] * 0.01
    row['b_power'] = row['b_accuracy'] * (1 - row['a_keeper']) + row['b_exp'] * 0.01
    X_new = pd.DataFrame([row])[X.columns]
    proba = model.predict_proba(X_new)[0]
    return {'A胜概率': round(proba[1], 3), 'B胜概率': round(proba[0], 3)}
# 案例：阿根廷 vs 法国
argentina = {'accuracy': 0.85, 'keeper': 0.30, 'exp': 9, 'is_first': True}
france    = {'accuracy': 0.88, 'keeper': 0.28, 'exp': 8, 'is_first': False}
result = predict_penalty_shootout(argentina, france)
print("阿根廷 vs 法国 点球预测：", result)

蒙特卡洛模拟（更直观）

def simulate_shootout(a_acc, b_acc, a_keeper, b_keeper,
                      a_first=True, n_sim=10000):
    """纯模拟点球大战过程"""
    a_wins = 0
    for _ in range(n_sim):
        sa = sb = 0
        # 5 轮常规
        for r in range(5):
            pressure = 0.9 if r == 4 else 1.0
            a_shoot = np.random.rand() < a_acc * (1 - b_keeper) * pressure
            b_shoot = np.random.rand() < b_acc * (1 - a_keeper) * pressure
            if a_first:
                sa += a_shoot
                sb += b_shoot
            else:
                sb += b_shoot
                sa += a_shoot
        # 突然死亡
        while sa == sb:
            a_ok = np.random.rand() < a_acc * (1 - b_keeper) * 0.85
            b_ok = np.random.rand() < b_acc * (1 - a_keeper) * 0.85
            sa += a_ok
            sb += b_ok
        if sa > sb:
            a_wins += 1
    return a_wins / n_sim
win_rate = simulate_shootout(0.85, 0.88, 0.30, 0.28, a_first=True)
print(f"模拟10000次，A胜率 = {win_rate:.2%}")