使用Pandas itertuples构建矩阵：填充每行指定日期后首次事件日期

原代码的逻辑仅判断当前行事件类型是否匹配，直接填充当前行日期或'na'，完全未实现当前行stamp日期之后和首次出现的核心需求，以下是修正后的实现方案：

高效实现方案（推荐） #

利用日期预处理+二分查找提升效率，避免逐行遍历全量数据：

import pandas as pd
from bisect import bisect_right

# 初始化原始数据
id_list =  [1,1,1,1,1,2,2,2,2,3,3,3,3,4,5,6,7,7,7,8,8,8,9,9,9,10,10,10]
fact_list = ["IC", "AC","IC","AC","IC","AC", "CC", "CD","IC","CC", "CD","IC","AC", "CD","IC","AC", "CC", "CD","IC","AC", "CC", "CD","IC","AC","IC","IC","AC","IC"]
stamp_list = ['1979-02-22','1973-11-06','1986-03-12','1986-01-24', '2012-05-22', '2009-01-18', '1992-01-14', '1985-06-05','2001-07-05','2008-11-19','2000-10-13','2002-04-18','1987-08-17','1977-04-09','1984-03-22','1994-08-08','2005-07-09','1982-05-03','2016-01-30','2019-03-10','1981-03-23','1979-07-21','2023-01-14','2018-06-23','1995-08-27','2020-11-08','2014-02-17','1977-09-08']
data = pd.DataFrame({"ID": id_list, "fact": fact_list, "stamp": stamp_list})

# 转换日期格式并排序，确保日期比较的正确性
data['stamp'] = pd.to_datetime(data['stamp'])
data.sort_values(by="stamp", inplace=True)

# 预处理：按事件类型分组，收集并排序所有对应日期
fact_dates = {}
for fact_type in ['AC', 'CD', 'IC', 'CC']:
    dates = data[data['fact'] == fact_type]['stamp'].sort_values().tolist()
    fact_dates[fact_type] = dates

# 定义函数：查找当前日期之后的首个事件日期
def get_first_future_date(current_stamp, date_list):
    idx = bisect_right(date_list, current_stamp)
    return date_list[idx] if idx < len(date_list) else pd.NaT

# 为每行生成各事件类型的首次未来日期列
for fact_type in ['AC', 'CD', 'IC', 'CC']:
    data[fact_type] = data['stamp'].apply(lambda x: get_first_future_date(x, fact_dates[fact_type]))

print(data)

逻辑说明 #

1. 先将stamp转为datetime类型，避免字符串比较的误差，同时对DataFrame按日期排序。
2. 预处理阶段把每个事件类型的所有日期收集并排序，用bisect_right快速定位第一个大于当前日期的位置，比逐行遍历全量数据效率提升显著。
3. 对每行的日期，调用函数匹配各事件类型的首个未来日期，无符合条件的日期返回pd.NaT（日期类型的缺失值）。

基于itertuples的修正版本 #

如果坚持使用itertuples遍历，可按以下方式修改：

import pandas as pd
from bisect import bisect_right

# 初始化并处理数据（同上方）
id_list =  [1,1,1,1,1,2,2,2,2,3,3,3,3,4,5,6,7,7,7,8,8,8,9,9,9,10,10,10]
fact_list = ["IC", "AC","IC","AC","IC","AC", "CC", "CD","IC","CC", "CD","IC","AC", "CD","IC","AC", "CC", "CD","IC","AC", "CC", "CD","IC","AC","IC","IC","AC","IC"]
stamp_list = ['1979-02-22','1973-11-06','1986-03-12','1986-01-24', '2012-05-22', '2009-01-18', '1992-01-14', '1985-06-05','2001-07-05','2008-11-19','2000-10-13','2002-04-18','1987-08-17','1977-04-09','1984-03-22','1994-08-08','2005-07-09','1982-05-03','2016-01-30','2019-03-10','1981-03-23','1979-07-21','2023-01-14','2018-06-23','1995-08-27','2020-11-08','2014-02-17','1977-09-08']
data = pd.DataFrame({"ID": id_list, "fact": fact_list, "stamp": stamp_list})
data['stamp'] = pd.to_datetime(data['stamp'])
data.sort_values(by="stamp", inplace=True)

# 准备结果结构
structure =  {'ID': [], 'stamp':[], 'fact': [], 'AC':[], 'CD':[], 'IC':[], 'CC':[]}
fact_types = ['AC', 'CD', 'IC', 'CC']

# 预处理每个事件的日期列表
fact_dates = {ft: data[data['fact'] == ft]['stamp'].sort_values().tolist() for ft in fact_types}

for row in data.itertuples():
    current_stamp = row.stamp
    # 填充基础字段
    structure["ID"].append(row.ID)
    structure["stamp"].append(current_stamp)
    structure["fact"].append(row.fact)
    
    # 为每个事件类型查找首个未来日期
    for ft in fact_types:
        dates = fact_dates[ft]
        idx = bisect_right(dates, current_stamp)
        structure[ft].append(dates[idx] if idx < len(dates) else pd.NaT)

# 转为DataFrame
result_df = pd.DataFrame(structure)
print(result_df)

内容的提问来源于stack exchange，提问作者ZuZiTeK