python切割时间数据_python – 如何切割具有日期字段作为索引的数据帧？

在我的数据框中,我将索引更改为日期字段

df.index = df.TRX_DATE # transaction date and type is class pandas.core.series.Series'

现在我想根据两个日期或任何日期差异来切割我的数据帧.

但我得到错误.

# currentdate is today date

startdate = currentdate - timedelta(days=30)

dflast30 = df.loc[startdate:currentdate] # error

尝试通过创建面具来做

mask = (df['TRX_DATE'] >= startdate) & (df['TRX_DATE'] <= currentdate )

dflast30 = df.loc[mask]

dflast30 = df.loc[mask]

TypeError: unorderable types: str() > datetime.datetime()

然后我尝试截断像：

dflast30 = df.truncate(before = currentdate, after = startdate)

我得到了同样的错误.

我很迷惑.我需要就这些问题提出建议：

>我可以将索引(TRX_DATE字段)更改为datetime类型吗？

>或者我应该创建该字符串字段类型.

>或者我应该按原样放置未分配的索引,并在日期字段中搜索我当前的要求.

>或者举一个例子,我如何将日期字段作为日期范围的索引和切片,请同时提及输出.

解决方法:

我认为你的第一种方法很好.

如果您希望复制列TRX_DATE索引：

df.index = pd.to_datetime(df['TRX_DATE'])

如果您不想复制,只需将列TRX_DATE设置为索引：

df = df.set_index(['TRX_DATE'])

有我的演示：

import pandas as pd

import numpy as np

import io

import datetime as dt

temp=u"""TRX_DATE;A

2013-07-05;1

2013-08-06;1

2015-09-05;2

2015-10-08;2

2015-11-05;2

2015-11-25;2

2015-12-06;3"""

df = pd.read_csv(io.StringIO(temp), sep=";", parse_dates=[0])

print df

# TRX_DATE A

#0 2013-07-05 1

#1 2013-08-06 1

#2 2015-09-05 2

#3 2015-10-08 2

#4 2015-11-05 2

#5 2015-11-25 2

#6 2015-12-06 3

print df.dtypes

#TRX_DATE datetime64[ns]

#A int64

#dtype: object

#copy column TRX_DATE to index

#df.index = pd.to_datetime(df['TRX_DATE'])

#no copy, only set column TRX_DATE to index

df = df.set_index(['TRX_DATE'])

print df

# A

#TRX_DATE

#2013-07-05 1

#2013-08-06 1

#2015-09-05 2

#2015-10-08 2

#2015-11-05 2

#2015-11-25 2

#2015-12-06 3

currentdate = dt.date.today()

print currentdate

#2015-11-06

startdate = currentdate - pd.Timedelta(days=30)

print startdate

#2015-10-07

dflast30 = df.loc[startdate:currentdate]

print dflast30

# A

#TRX_DATE

#2015-10-08 2

#2015-11-05 2

dflast30 = dflast30.reset_index()

print dflast30

# TRX_DATE A

#0 2015-10-08 2

#1 2015-11-05 2

不同的方法,您可以在其中创建df的子集.无需设置datetimeindex.

import pandas as pd

import numpy as np

import io

import datetime as dt

temp=u"""TRX_DATE;A

2013-07-05;1

2013-08-06;1

2015-09-05;2

2015-10-08;2

2015-11-05;2

2015-11-25;2

2015-12-06;3"""

df = pd.read_csv(io.StringIO(temp), sep=";", parse_dates=[0])

print df

# TRX_DATE A

#0 2013-07-05 1

#1 2013-08-06 1

#2 2015-09-05 2

#3 2015-10-08 2

#4 2015-11-05 2

#5 2015-11-25 2

#6 2015-12-06 3

print df.dtypes

#TRX_DATE datetime64[ns]

#A int64

#dtype: object

currentdate = dt.date.today()

print currentdate

#2015-11-06

startdate = currentdate - pd.Timedelta(days=30)

print startdate

#2015-10-07

dflast30 = df[(df.TRX_DATE >= startdate) & (df.TRX_DATE <= currentdate)]

print dflast30

# TRX_DATE A

#3 2015-10-08 2

#4 2015-11-05 2

标签：python,date,datetime,pandas

来源： https://codeday.me/bug/20190702/1357214.html