我有以下数据框df：
Opname_start_date Opname_end_date
0 2016-11-21 11:33:00 2016-11-23 12:02:59.0000000
1 2016-07-06 09:16:00 2016-07-06 09:16:01.0000000
2 2016-08-11 10:18:00 2016-08-15 10:40:59.0000000
3 2016-09-23 11:28:00 2016-09-23 11:28:01.0000000
4 2016-08-11 14:06:00 2016-08-11 14:06:01.0000000
5 2016-10-26 09:42:00 2016-10-29 11:25:59.0000000
6 2016-12-02 15:03:00 2016-12-09 14:00:00.0000000
7 2016-07-08 09:00:00 2016-09-27 09:15:01.0000000
8 2016-06-14 09:01:00 2016-06-17 13:00:00.0000000
9 2016-12-08 13:54:00 2016-12-08 13:54:01.0000000
我想从Opname_start_date中减去Opname_end_date，并将其保存到新列中。我尝试了以下方法：
df['LOS'] = df['Opname_end_date'] - df['Opname_start_date']
但是它给了我以下错误：
TypeError: ufunc subtract cannot use operands with types dtype('<U27') and dtype('<M8[ns]')
我检查了数据帧的dtypes，它显示第一列是datetime64 [ns]，但是第二列仍然是object。我尝试使用pd.to_datetime(df['Opname_end_date'])转换第二列，但它仍显示该列属于object类型，并且不会引发任何错误。
我无法弄清楚问题出在哪里。如果有人可以帮助我，那就太好了。谢谢
答案 0 :(得分：2)
对我来说，to_datetime会将列Opname_end_date转换为datetimes：
print (df.dtypes)
Opname_start_date datetime64[ns]
Opname_end_date object
dtype: object
df['LOS'] = pd.to_datetime(df['Opname_end_date']) - df['Opname_start_date']
或者：
df['Opname_end_date'] = pd.to_datetime(df['Opname_end_date'])
df['LOS'] = df['Opname_end_date'] - df['Opname_start_date']
print (df)
Opname_start_date Opname_end_date LOS
0 2016-11-21 11:33:00 2016-11-23 12:02:59.0000000 2 days 00:29:59
1 2016-07-06 09:16:00 2016-07-06 09:16:01.0000000 0 days 00:00:01
2 2016-08-11 10:18:00 2016-08-15 10:40:59.0000000 4 days 00:22:59
3 2016-09-23 11:28:00 2016-09-23 11:28:01.0000000 0 days 00:00:01
4 2016-08-11 14:06:00 2016-08-11 14:06:01.0000000 0 days 00:00:01
5 2016-10-26 09:42:00 2016-10-29 11:25:59.0000000 3 days 01:43:59
6 2016-12-02 15:03:00 2016-12-09 14:00:00.0000000 6 days 22:57:00
7 2016-07-08 09:00:00 2016-09-27 09:15:01.0000000 81 days 00:15:01
8 2016-06-14 09:01:00 2016-06-17 13:00:00.0000000 3 days 03:59:00
9 2016-12-08 13:54:00 2016-12-08 13:54:01.0000000 0 days 00:00:01