python - 根据分组日期值更新列

Question

使用正确的示例输出编辑/重新发布。

我有一个如下所示的数据框:

data = {
    "ID": [1, 1, 1, 2, 2, 2],
  "Year": [2021, 2021, 2023, 2015, 2017, 2018],
  "Combined": ['started', 'finished', 'started', 'started', 'finished', 'started'],
  "bool": [True, False, False, True, False, False],
"Update": ['started', 'finished', 'started', 'started', 'finished', 'started']

}

df = pd.DataFrame(data)

print(df) 

ID  Year  Combined   bool  
1  2021   started   True   
1  2021  finished  False  
1  2023   started  False   
2  2015   started   True   
2  2017  finished  False  
2  2018   started  False   

此数据框按 ID 分为几组。

我想基于 if df['bool'] == True 创建更新的combined 列，但前提是 df['bool' ] == True 并且同一组中还有另一个“已完成”行，其年份较晚(不相同)。

示例输出:

ID  Year  Combined   bool    Update
1  2021   started   True   started
1  2021  finished  False  finished
1  2023   started  False   started
2  2015   started   True   finished
2  2017  finished  False  finished
2  2018   started  False   started

我们不会更新第一组，因为在较晚的一年中没有 finished 值，而我们正在更新第二组，因为在稍后的一年。谢谢!

Answer 1

我能想到的一个解决方案是使用 apply 和 groupby 方法。每个组的值通过 update 函数传递给 update_group 函数。这允许执行测试并在满足条件时返回更新的“Update”列。然后，返回的 DataFrame 就是预期的 DataFrame。

如果我接管您的示例，则省略将在第二部分中创建的“Update”列:

import pandas as pd

data = {
    "ID": [1, 1, 1, 2, 2, 2],
    "Year": [2021, 2021, 2023, 2015, 2017, 2018],
    "Combined": ["started", 'finished', 'started', 'started', 'finished', 'started'],
    "bool": [True, False, False, True, False, False],
}

df = pd.DataFrame(data)
print(df)

我获得以下输入数据帧:

   ID  Year  Combined   bool
0   1  2021   started   True
1   1  2021  finished  False
2   1  2023   started  False
3   2  2015   started   True
4   2  2017  finished  False
5   2  2018   started  False

这是我用来更新 DataFrame 的两个函数:

def update_group(row, group):
    """Update each row of a group"""
    if row["bool"] is True:
        # Extract the later years entries
        group_later = group[group.Year > row.Year]

        # If finished in found, then turn the Update column to finished
        if any(group_later.Combined == "finished"):
            row["Update"] = "finished"
        else:

            row["Update"] = row["Combined"]
    else:
        row["Update"] = row["Combined"]

    return row

def update(group):
    """Apply the update to each group"""
    return group.apply(update_group, group=group, axis=1)

因此，如果您将这些函数应用到 DataFrame 中:

df = df.groupby("ID").apply(update)
print(df)

返回的DataFrame是:

   ID  Year  Combined   bool    Update
0   1  2021   started   True   started
1   1  2021  finished  False  finished
2   1  2023   started  False   started
3   2  2015   started   True  finished
4   2  2017  finished  False  finished
5   2  2018   started  False   started