Skip to main content
 首页 » 编程设计

python之使用 pandas read_csv 解析时间戳时出错

2025年04月02日10Terrylee

我正在尝试加载格式如下的 csv 文件:

                  40010  40015  40020  40025  40030  40035  40040  40045   
2008-11-03 00:00    786    212    779    227    220    131    680   1006    
2008-11-03 00:03    760    200    765    234    225    133    694   1063    
2008-11-03 00:06    757    205    769    237    230    136    726   1051    
2008-11-03 00:09    781    207    765    240    235    137    711   1040    
2008-11-03 00:12    759    203    751    232    225    134    717   1088 
... 

文件以逗号分隔。这里没有固定宽度。

我希望行索引是日期时间,所以这是我当前在加载文件时正在做的事情:

def dateparse (timestamp):    
    return datetime.datetime.strptime(timestamp, '%Y-%m-%d %I:%M') 
 
global_data_train = pd.read_csv('RTAHistorical.csv', sep=",",parse_dates=True, date_parser=dateparse, header=0, index_col=0, skip_blank_lines = True, engine='python') 

但我收到以下错误:

TypeError: strptime() argument 1 must be str, not numpy.ndarray 

如我所见some people成功地使用了相同的方法,我不太明白这个错误。

我做错了什么?

请您参考如下方法:

对我来说,将格式更改为 %Y-%m-%d %H:%M:

def dateparse (timestamp):    
    return pd.datetime.strptime(timestamp, '%Y-%m-%d %H:%M') 

示例:

import pandas as pd 
from pandas.compat import StringIO 
 
temp=u"""40010,40015,40020,40025,40030,40035,40040,40045 
2008-11-03 00:00,786,212,779,227,220,131,680,1006 
2008-11-03 00:03,760,200,765,234,225,133,694,1063 
2008-11-03 00:06,757,205,769,237,230,136,726,1051 
2008-11-03 00:09,781,207,765,240,235,137,711,1040 
2008-11-03 00:12,759,203,751,232,225,134,717,1088""" 
#after testing replace StringIO(temp) to filename 
def dateparse (timestamp):    
    return pd.datetime.strptime(timestamp, '%Y-%m-%d %H:%M') 
 
global_data_train = pd.read_csv(StringIO(temp),  
                                sep=",",  
                                parse_dates=True,  
                                date_parser=dateparse,  
                                header=0,  
                                index_col=0,  
                                skip_blank_lines = True,  
                                engine='python') 
print (global_data_train) 
                     40010  40015  40020  40025  40030  40035  40040  40045 
2008-11-03 00:00:00    786    212    779    227    220    131    680   1006 
2008-11-03 00:03:00    760    200    765    234    225    133    694   1063 
2008-11-03 00:06:00    757    205    769    237    230    136    726   1051 
2008-11-03 00:09:00    781    207    765    240    235    137    711   1040 
2008-11-03 00:12:00    759    203    751    232    225    134    717   1088 
 
print (global_data_train.index) 
DatetimeIndex(['2008-11-03 00:00:00', '2008-11-03 00:03:00', 
               '2008-11-03 00:06:00', '2008-11-03 00:09:00', 
               '2008-11-03 00:12:00'], 
              dtype='datetime64[ns]', freq=None) 

也可以省略 date_parser=dateparse。

import pandas as pd 
from pandas.compat import StringIO 
 
temp=u"""40010,40015,40020,40025,40030,40035,40040,40045 
2008-11-03 00:00,786,212,779,227,220,131,680,1006 
2008-11-03 00:03,760,200,765,234,225,133,694,1063 
2008-11-03 00:06,757,205,769,237,230,136,726,1051 
2008-11-03 00:09,781,207,765,240,235,137,711,1040 
2008-11-03 00:12,759,203,751,232,225,134,717,1088""" 
#after testing replace StringIO(temp) to filename 
global_data_train = pd.read_csv(StringIO(temp),  
                                parse_dates=True,  
                                skip_blank_lines = True) 
print (global_data_train) 
                     40010  40015  40020  40025  40030  40035  40040  40045 
2008-11-03 00:00:00    786    212    779    227    220    131    680   1006 
2008-11-03 00:03:00    760    200    765    234    225    133    694   1063 
2008-11-03 00:06:00    757    205    769    237    230    136    726   1051 
2008-11-03 00:09:00    781    207    765    240    235    137    711   1040 
2008-11-03 00:12:00    759    203    751    232    225    134    717   1088 
 
print (global_data_train.index) 
DatetimeIndex(['2008-11-03 00:00:00', '2008-11-03 00:03:00', 
               '2008-11-03 00:06:00', '2008-11-03 00:09:00', 
               '2008-11-03 00:12:00'], 
              dtype='datetime64[ns]', freq=None)