python案例统计界外球进攻威胁次数？

统计界外球进攻威胁次数

下面是一个完整的 Python 案例，演示如何统计界外球（Throw-in）后的进攻威胁次数，核心思路是：找到界外球事件 → 追踪随后 N 秒/若干次传球内是否产生射门/射正/进球/进入禁区等威胁事件。

数据准备

假设你有事件数据（类似 StatsBomb / Opta 格式）,字段如下：

威胁定义（可自定义）

一次界外球进攻被计为有威胁，需在界外球后 10 秒 + 最多 3 次传球内出现以下任一：

• 射门 Shot（含射正 Shot on Target）
• 进球 Goal
• 传球进入对方禁区（x ≥ 83，y 在 18–82，按 0-100 坐标）
• 传中进入小禁区（x ≥ 94）

完整代码

import pandas as pd
THREAT_WINDOW_SEC = 10      # 界外球后观察窗口（秒）
MAX_PASS_CHAIN   = 3        # 最多追踪传球次数
BOX_X_START      = 83       # 对方禁区起点 x
BOX_Y_MIN, BOX_Y_MAX = 18, 82
SIX_X_START      = 94       # 小禁区 x
THREAT_TYPES = {"Shot", "Shot on Target", "Goal"}
def is_into_box(row):
    """传球终点是否进入对方禁区"""
    if pd.isna(row.get("end_x")):
        return False
    return row["end_x"] >= BOX_X_START and BOX_Y_MIN <= row["end_y"] <= BOX_Y_MAX
def is_into_six_yard(row):
    return (not pd.isna(row.get("end_x"))) and row["end_x"] >= SIX_X_START
def count_threats_from_throwins(df: pd.DataFrame) -> pd.DataFrame:
    """
    返回每场比赛的界外球总数与威胁界外球数
    """
    df = df.sort_values(["match_id", "period", "timestamp"]).reset_index(drop=True)
    results = []
    for match_id, g in df.groupby("match_id"):
        g = g.reset_index(drop=True)
        throw_idx = g.index[g["type"] == "Throw-in"].tolist()
        total, threat = 0, 0
        for i in throw_idx:
            total += 1
            throw_row = g.loc[i]
            team = throw_row["team"]
            t0 = throw_row["timestamp"]
            period = throw_row["period"]
            # 进入后续事件窗口
            j = i + 1
            pass_count = 0
            threatened = False
            while j < len(g):
                row = g.loc[j]
                # 换半场/换队（非本队事件）即中断链条
                if row["period"] != period:
                    break
                if row["timestamp"] - t0 > THREAT_WINDOW_SEC:
                    break
                if row["team"] != team:      # 球权丢失，链条终止
                    break
                etype = row["type"]
                # 威胁判定
                if etype in THREAT_TYPES:
                    threatened = True
                    break
                if etype == "Pass":
                    if is_into_box(row) or is_into_six_yard(row):
                        threatened = True
                        break
                    pass_count += 1
                    if pass_count > MAX_PASS_CHAIN:
                        break
                j += 1
            if threatened:
                threat += 1
        results.append({
            "match_id": match_id,
            "total_throwins": total,
            "threat_throwins": threat,
            "threat_rate": round(threat / total, 3) if total else 0
        })
    return pd.DataFrame(results)
# ---------- 构造示例数据 ----------
data = [
    # 比赛1：一次界外球后 5 秒射门（威胁）
    {"match_id": 1, "period": 1, "timestamp": 100, "team": "A", "type": "Throw-in", "x": 40, "y": 50, "end_x": None, "end_y": None},
    {"match_id": 1, "period": 1, "timestamp": 103, "team": "A", "type": "Pass",     "x": 45, "y": 55, "end_x": 90, "end_y": 40},
    {"match_id": 1, "period": 1, "timestamp": 105, "team": "A", "type": "Shot",     "x": 90, "y": 40, "end_x": None, "end_y": None},
    # 比赛1：一次界外球后无威胁（3次传球内没进禁区）
    {"match_id": 1, "period": 1, "timestamp": 200, "team": "A", "type": "Throw-in", "x": 30, "y": 50, "end_x": None, "end_y": None},
    {"match_id": 1, "period": 1, "timestamp": 203, "team": "A", "type": "Pass",     "x": 35, "y": 55, "end_x": 50, "end_y": 60},
    {"match_id": 1, "period": 1, "timestamp": 206, "team": "A", "type": "Pass",     "x": 55, "y": 60, "end_x": 70, "end_y": 65},
    # 比赛2：界外球后传球直达禁区（威胁）
    {"match_id": 2, "period": 1, "timestamp": 50, "team": "B", "type": "Throw-in",  "x": 60, "y": 20, "end_x": None, "end_y": None},
    {"match_id": 2, "period": 1, "timestamp": 53, "team": "B", "type": "Pass",      "x": 65, "y": 25, "end_x": 85, "end_y": 50},
]
df = pd.DataFrame(data)
result = count_threats_from_throwins(df)
print(result)

输出：

   match_id  total_throwins  threat_throwins  threat_rate
0         1               2                1          0.5
1         2               1                1          1.0

关键设计说明

进阶扩展

1. 使用 xG 而非二分：把窗口内所有射门的 xG 累加，作为"威胁强度"更细腻。

   xg_threat = g.loc[i+1:j, "xg"].sum()

2. 区分界外球位置：
   • 前场界外球（x ≥ 60）vs 后场界外球
   • 左侧/右侧
3. Defensive Third / Middle / Final Third 分区统计 threat_rate。
4. 可视化：用 matplotlib 画界外球起点散点图,颜色表示是否威胁。

如果你能提供实际数据结构（列名/示例行），我可以把上面的代码直接改写成适配你数据的版本（StatsBomb 的 event.json 或 Wyscout 格式）。