python - 在python中将csv对象时间解析为日期时间

Question

我有一个 csv 文件，下面有 Timestamp 列。我想将格式更改为 2013-08-12 10:29:19.673 或一秒的粒度。当前Timestamp 是object 类型。

我可以手动更改 Excel 中的格式，但文件太大，有些行会丢失。

        Id          Timestamp Data  Group_Id
0       19929927    00:07.5   27.0  27
1       19929928    00:08.3   26.5  27
2       19929929    00:48.7   33.5  157
3       19929930    00:50.0   33.0  157
4       19929931    00:53.1   35.0  25

                 ...

1048570 20978497    10:11.9   34.5  152
1048571 20978498    10:13.3   34.0  152
1048572 20978499    10:41.2   42.0  138
1048573 20978500    10:42.5   45.0  138
1048574 20978501    10:43.9   44.0  138

Answer 1

编辑:如果将时间转换为没有日期信息的日期时间，pandas 显然会添加实际日期的日期。

如果还需要几天，请检查此解决方案:

想法是如果时间以 0 开头，则创建连续的日期时间组合:

df = df[['Timestamp']]
print (df)
   Timestamp
0    00:08.3 <- first day
1    00:48.7
2    00:50.0
3    00:53.1
4    10:11.9
5    10:13.3
6    10:41.2
7    00:50.0 <- second day
8    00:53.1
9    10:42.5
10   10:43.9
11   00:07.5 <- third day
12   00:08.3
13   10:11.9
14   10:13.3
15   10:43.9

<小时/>

#convert to datetimes and get hours for test 0
df['h'] = pd.to_datetime(df['Timestamp']).dt.hour
#test first 0 for start of day
df['mask'] = df['h'].shift().ne(0) & df['h'].eq(0)
#create consecutive groups - starts by 1 if first time start by 0, else start by 1 
df['g'] = df['mask'].cumsum()
#specify first day in origin parameter
df['days'] = pd.to_datetime(df['g'], origin='2016-01-01', unit='d')
#add to original Timestamps if HH:MM.SS
df['Timestamp1'] = df['days'] + pd.to_timedelta(df['Timestamp'].str.replace('\.',':'))
#add to original Timestamps if format without hours - MM:SS.SS
df['Timestamp2'] = df['days'] + pd.to_timedelta('00:' + df['Timestamp'])

<小时/>

print (df)
   Timestamp   h   mask  g       days          Timestamp1  \
0    00:08.3   0   True  1 2016-01-02 2016-01-02 00:08:03   
1    00:48.7   0  False  1 2016-01-02 2016-01-02 00:48:07   
2    00:50.0   0  False  1 2016-01-02 2016-01-02 00:50:00   
3    00:53.1   0  False  1 2016-01-02 2016-01-02 00:53:01   
4    10:11.9  10  False  1 2016-01-02 2016-01-02 10:11:09   
5    10:13.3  10  False  1 2016-01-02 2016-01-02 10:13:03   
6    10:41.2  10  False  1 2016-01-02 2016-01-02 10:41:02   
7    00:50.0   0   True  2 2016-01-03 2016-01-03 00:50:00   
8    00:53.1   0  False  2 2016-01-03 2016-01-03 00:53:01   
9    10:42.5  10  False  2 2016-01-03 2016-01-03 10:42:05   
10   10:43.9  10  False  2 2016-01-03 2016-01-03 10:43:09   
11   00:07.5   0   True  3 2016-01-04 2016-01-04 00:07:05   
12   00:08.3   0  False  3 2016-01-04 2016-01-04 00:08:03   
13   10:11.9  10  False  3 2016-01-04 2016-01-04 10:11:09   
14   10:13.3  10  False  3 2016-01-04 2016-01-04 10:13:03   
15   10:43.9  10  False  3 2016-01-04 2016-01-04 10:43:09   

                Timestamp2  
0  2016-01-02 00:00:08.300  
1  2016-01-02 00:00:48.700  
2  2016-01-02 00:00:50.000  
3  2016-01-02 00:00:53.100  
4  2016-01-02 00:10:11.900  
5  2016-01-02 00:10:13.300  
6  2016-01-02 00:10:41.200  
7  2016-01-03 00:00:50.000  
8  2016-01-03 00:00:53.100  
9  2016-01-03 00:10:42.500  
10 2016-01-03 00:10:43.900  
11 2016-01-04 00:00:07.500  
12 2016-01-04 00:00:08.300  
13 2016-01-04 00:10:11.900  
14 2016-01-04 00:10:13.300  
15 2016-01-04 00:10:43.900