Question

我正在尝试将Datetime转换为Unix时间纪元，但我收到了以下错误。

输入：

userid,datetime,latitude,longitude
156,2014-02-01 00:00:00.739166+01,41.8836718276551,12.4877775603346
187,2014-02-01 00:00:01.148457+01,41.9285433333333,12.4690366666667
297,2014-02-01 00:00:01.220066+01,41.8910686119733,12.4927045625339
89,2014-02-01 00:00:01.470854+01,41.7931766914244,12.4321219603157
79,2014-02-01 00:00:01.631136+01,41.90027472,12.46274618
191,2014-02-01 00:00:02.048546+01,41.8523047579646,12.5774065771898
343,2014-02-01 00:00:02.647839+01,41.8921718255185,12.4696996165151
341,2014-02-01 00:00:02.709888+01,41.9102125627332,12.4770004336041
260,2014-02-01 00:00:03.458195+01,41.8658208551143,12.4655221109313

程序：

import pandas as pd
import numpy as np
import io

df = pd.read_csv('input.csv', 
                 #header=None, #no header in csv
                 header=['userid','datetime','latitude','longitude'], #set custom column names
                 parse_dates=['datetime']) #parse columns d, e to datetime

df['datetime'] = df['datetime'].astype(np.int64) // 10**9
#df['e'] = df['e'].astype(np.int64) // 10**9

df.to_csv('output.csv', header=True, index=False)

上面的程序在python 2.7中工作正常但不是我升级到python 3.x Anaconda我无法获得结果

错误：

  File "pandas\parser.pyx", line 519, in pandas.parser.TextReader.__cinit__ (pandas\parser.c:5907)

TypeError: Can't convert 'int' object to str implicitly

修改：输入文件here

Answer 1

如果csv没有标头，则必须参数listExtendQuery和names parse_dates - 尝试将第二列解析为[1]：

datetime

import pandas as pd
import numpy as np
from pandas.compat import StringIO

temp=u"""156,2014-02-01 00:00:00.739166+01,41.8836718276551,12.4877775603346
187,1014-02-01 00:00:01.148457+01,41.9285433333333,12.4690366666667
297,2014-02-01 00:00:01.220066+01,41.8910686119733,12.4927045625339
89,2014-02-01 00:00:01.470854+01,41.7931766914244,12.4321219603157
79,2014-02-01 00:00:01.631136+01,41.90027472,12.46274618
191,2014-02-01 00:00:02.048546+01,41.8523047579646,12.5774065771898
343,2014-02-01 00:00:02.647839+01,41.8921718255185,12.4696996165151
341,2014-02-01 00:00:02.709888+01,41.9102125627332,12.4770004336041
260,2014-02-01 00:00:03.458195+01,41.8658208551143,12.4655221109313"""
#after testing replace 'StringIO(temp)' to 'filename.csv'
df = pd.read_csv(StringIO(temp), 
                parse_dates=[1], 
                names=['userid','datetime','latitude','longitude'])
#print (df)

#check dtypes if datetime it is OK
print (df['datetime'].dtypes)
datetime64[ns]

另一个可能的问题是错误的数据，在我的第二行示例中：

df['datetime'] = df['datetime'].astype(np.int64) // 10**9
print (df)
   userid    datetime   latitude  longitude
0     156  1391209200  41.883672  12.487778
1     187  1391209201  41.928543  12.469037
2     297  1391209201  41.891069  12.492705
3      89  1391209201  41.793177  12.432122
4      79  1391209201  41.900275  12.462746
5     191  1391209202  41.852305  12.577407
6     343  1391209202  41.892172  12.469700
7     341  1391209202  41.910213  12.477000
8     260  1391209203  41.865821  12.465522

使用to_datetime和参数import pandas as pd from pandas.compat import StringIO temp=u"""156,2014-02-01 00:00:00.739166+01,41.8836718276551,12.4877775603346 187,1014-02-01 00:00:01.148457+01,41.9285433333333,12.4690366666667 297,2014-02-01 00:00:01.220066+01,41.8910686119733,12.4927045625339 89,2014-02-01 00:00:01.470854+01,41.7931766914244,12.4321219603157 79,2014-02-01 00:00:01.631136+01,41.90027472,12.46274618 191,2014-02-01 00:00:02.048546+01,41.8523047579646,12.5774065771898 343,2014-02-01 00:00:02.647839+01,41.8921718255185,12.4696996165151 341,2014-02-01 00:00:02.709888+01,41.9102125627332,12.4770004336041 260,2014-02-01 00:00:03.458195+01,41.8658208551143,12.4655221109313""" #after testing replace 'StringIO(temp)' to 'filename.csv' df = pd.read_csv(StringIO(temp), parse_dates=[1], names=['userid','datetime','latitude','longitude']) #print (df) #check dtypes - parse failed, get object dtype print (df['datetime'].dtypes) object解析到日期时间 - 它将错误数据替换为errors='coerce'，然后将NaT替换为某个值，例如NaT {（1}}）与fillna：

编辑：

如果还有标题且需要替换列名，则需要1970-01-01 00:00:00.000000添加到read_csv。

Answer 2

pd.read_csv中的header参数需要int或int列表而不是字符串列表。

from io import StringIO
file="""
userid,datetime,latitude,longitude
156,2014-02-01 00:00:00.739166+01,41.8836718276551,12.4877775603346
187,2014-02-01 00:00:01.148457+01,41.9285433333333,12.4690366666667
297,2014-02-01 00:00:01.220066+01,41.8910686119733,12.4927045625339
89,2014-02-01 00:00:01.470854+01,41.7931766914244,12.4321219603157
79,2014-02-01 00:00:01.631136+01,41.90027472,12.46274618
191,2014-02-01 00:00:02.048546+01,41.8523047579646,12.5774065771898
343,2014-02-01 00:00:02.647839+01,41.8921718255185,12.4696996165151
341,2014-02-01 00:00:02.709888+01,41.9102125627332,12.4770004336041
260,2014-02-01 00:00:03.458195+01,41.8658208551143,12.4655221109313"""

让我们试试这个read_csv语句：

df = pd.read_csv(StringIO(file),parse_dates=['datetime'])
df['datetime'] = df['datetime'].astype(np.int64) // 10**9

print(df.head())

输出：

   userid    datetime   latitude  longitude
0     156  1391209200  41.883672  12.487778
1     187  1391209201  41.928543  12.469037
2     297  1391209201  41.891069  12.492705
3      89  1391209201  41.793177  12.432122
4      79  1391209201  41.900275  12.462746

int - 将datetime转换为Unix时间纪元时的字符串类型错误

2 个答案: