基于Python Pandas DataFrame中的条件遍历行和更新列值

由于您的情况不是那么简单，您需要reshape 您的数据帧，然后按Issue_Id分组，然后更新值.因为您的数据框是按"CHANGED_DATE"排序的，所以我们的 idea 是向前填充新值，向后填充旧值.如果缺少该参数，只需使用现有值填充即可:

def update_values(df):
    return df['new_value'].ffill().fillna(df['old_value'].bfill())
        
upd_values = (df.pivot_table(index=df.index, columns='changed_parameter',
                             values=['old_value', 'new_value'], aggfunc='first')
                .groupby(df['Issue_Id']).apply(update_values)
                .droplevel('Issue_Id').fillna(df))

df[upd_values.columns] = upd_values

输出:

>>> df
    Issue_Id   Due_Date      status  estimation_hour changed_date changed_parameter   old_value   new_value
0        101  1/20/2023    Accepted                0    1/10/2023            status     Defined    Accepted
1        101  1/20/2023    Accepted               20    1/15/2023   estimation_hour           0          20
2        101  1/20/2023    Accepted               30    1/16/2023   estimation_hour          20          30
3        101  1/31/2023    Accepted               30    1/16/2023          Due_Date   1/20/2023   1/31/2023
4        101  1/31/2023  InProgress               30    1/20/2023            status    Accepted  InProgress
5        101  1/31/2023  InProgress               40    1/25/2023   estimation_hour          30          40
6        101  1/31/2023      Closed               40    1/30/2023            status  InProgress      Closed
7        102  2/28/2023    Accepted                0    1/10/2023            status     Defined    Accepted
8        102  2/28/2023    Accepted               30    1/15/2023   estimation_hour           0          30
9        102  2/28/2023  InProgress               30    1/20/2023            status    Accepted  InProgress
10       102  2/28/2023  InProgress               50    1/25/2023   estimation_hour          30          50
11       102  2/28/2023      Closed               50    1/30/2023            status  InProgress      Closed

Old answer个

old_value = pd.to_numeric(df['old_value'], errors='coerce').shift(-1)
new_value = pd.to_numeric(df['new_value'], errors='coerce')

df['estimation_hour'] = old_value.fillna(new_value).ffill().convert_dtypes()
df['status'] = df['new_value'].mask(new_value.notna()).ffill()

输出:

>>> df
   Issue_Id      status  estimation_hour changed_date changed_parameter   old_value   new_value
0       101    Accepted                0    1/10/2023            status     Defined    Accepted
1       101    Accepted               20    1/15/2023   estimation_hour           0          20
2       101  InProgress               20    1/20/2023            status    Accepted  InProgress
3       101  InProgress               40    1/25/2023   estimation_hour          20          40
4       101      Closed               40    1/30/2023            status  InProgress      Closed
5       102    Accepted                0    1/10/2023            status     Defined    Accepted
6       102    Accepted               30    1/15/2023   estimation_hour           0          30
7       102  InProgress               30    1/20/2023            status    Accepted  InProgress
8       102  InProgress               50    1/25/2023   estimation_hour          30          50
9       102      Closed               50    1/30/2023            status  InProgress      Closed