미국 일광 절약 시간제로 발생하는 1 시간 시간 이동 문제를 해결하려고합니다. 내가 사용하여 새 날짜 범위를 적용 할 때 시계열의 일부 서머 타임이 1 시간 바뀔 때 잘못된 색인 다시 생성
이
현재In [3] eurusd
Out[3]:
BID-CLOSE
TIME
1994-03-28 22:00:00 1.15981
1994-03-29 22:00:00 1.16681
1994-03-30 22:00:00 1.15021
1994-03-31 22:00:00 1.14851
1994-04-03 21:00:00 1.14081
1994-04-04 21:00:00 1.13921
1994-04-05 21:00:00 1.13881
1994-04-06 21:00:00 1.14351
1994-04-07 21:00:00 1.14411
1994-04-10 21:00:00 1.14011
1994-04-11 21:00:00 1.14391
1994-04-12 21:00:00 1.14451
1994-04-13 21:00:00 1.14201
1994-04-14 21:00:00 1.13911
1994-04-17 21:00:00 1.14821
1994-04-18 21:00:00 1.15181
1994-04-19 21:00:00 1.15621
1994-04-20 21:00:00 1.15381
1994-04-21 21:00:00 1.16201
1994-04-24 21:00:00 1.16251
1994-04-25 21:00:00 1.16721
1994-04-26 21:00:00 1.17101
1994-04-27 21:00:00 1.17721
1994-04-28 21:00:00 1.18421
1994-05-01 21:00:00 1.18751
1994-05-02 21:00:00 1.17331
1994-05-03 21:00:00 1.16801
1994-05-04 21:00:00 1.17141
1994-05-05 21:00:00 1.17691
1994-05-08 21:00:00 1.16541
...
1994-09-26 21:00:00 1.25501
1994-09-27 21:00:00 1.25761
1994-09-28 21:00:00 1.25541
1994-09-29 21:00:00 1.25421
1994-10-02 21:00:00 1.25721
1994-10-03 21:00:00 1.26131
1994-10-04 21:00:00 1.26121
1994-10-05 21:00:00 1.26101
1994-10-06 21:00:00 1.25761
1994-10-10 21:00:00 1.26161
1994-10-11 21:00:00 1.26341
1994-10-12 21:00:00 1.27821
1994-10-13 21:00:00 1.29411
1994-10-16 21:00:00 1.29401
1994-10-17 21:00:00 1.29371
1994-10-18 21:00:00 1.29531
1994-10-19 21:00:00 1.29681
1994-10-20 21:00:00 1.29971
1994-10-23 21:00:00 1.30411
1994-10-24 21:00:00 1.30311
1994-10-25 21:00:00 1.30091
1994-10-26 21:00:00 1.28921
1994-10-27 21:00:00 1.29341
1994-10-30 22:00:00 1.29931
1994-10-31 22:00:00 1.29281
1994-11-01 22:00:00 1.27771
1994-11-02 22:00:00 1.27821
1994-11-03 22:00:00 1.28321
1994-11-06 22:00:00 1.28751
1994-11-07 22:00:00 1.27091
(아래 자르는 것은) : 다음
idx = pd.date_range('1994-03-28 22:00:00', '1994-11-07 22:00:00', freq= 'D')
In [4] idx
Out[4]:
DatetimeIndex(['1994-03-28 22:00:00', '1994-03-29 22:00:00',
'1994-03-30 22:00:00', '1994-03-31 22:00:00',
'1994-04-01 22:00:00', '1994-04-02 22:00:00',
'1994-04-03 22:00:00', '1994-04-04 22:00:00',
'1994-04-05 22:00:00', '1994-04-06 22:00:00',
...
'1994-10-29 22:00:00', '1994-10-30 22:00:00',
'1994-10-31 22:00:00', '1994-11-01 22:00:00',
'1994-11-02 22:00:00', '1994-11-03 22:00:00',
'1994-11-04 22:00:00', '1994-11-05 22:00:00',
'1994-11-06 22:00:00', '1994-11-07 22:00:00'],
dtype='datetime64[ns]', length=225, freq='D')
을 나는 새 날짜 범위를 사용하여 dataframe를 다시 인덱싱 timeseries는 모든 21:00 값을 22:00로 변환하고 BID-CLOSE는 NaN으로 변환합니다. 이유를 이해하지만, 코드를 US Summer Time 스케쥴에 따라 1 시간 단위로 인식하는 방법을 확신 할 수 없습니다. 인덱싱을
의출력 :
In[5]: eurusd_copy1 = eurusd.reindex(idx, fill_value=None)
In[6]: eurusd_copy1
Out[6]:
BID-CLOSE
1994-03-28 22:00:00 1.15981
1994-03-29 22:00:00 1.16681
1994-03-30 22:00:00 1.15021
1994-03-31 22:00:00 1.14851
1994-04-01 22:00:00 NaN
1994-04-02 22:00:00 NaN
1994-04-03 22:00:00 NaN
1994-04-04 22:00:00 NaN
1994-04-05 22:00:00 NaN
1994-04-06 22:00:00 NaN
1994-04-07 22:00:00 NaN
1994-04-08 22:00:00 NaN
1994-04-09 22:00:00 NaN
1994-04-10 22:00:00 NaN
1994-04-11 22:00:00 NaN
1994-04-12 22:00:00 NaN
1994-04-13 22:00:00 NaN
1994-04-14 22:00:00 NaN
1994-04-15 22:00:00 NaN
1994-04-16 22:00:00 NaN
1994-04-17 22:00:00 NaN
1994-04-18 22:00:00 NaN
1994-04-19 22:00:00 NaN
1994-04-20 22:00:00 NaN
1994-04-21 22:00:00 NaN
1994-04-22 22:00:00 NaN
1994-04-23 22:00:00 NaN
1994-04-24 22:00:00 NaN
1994-04-25 22:00:00 NaN
1994-04-26 22:00:00 NaN
...
1994-10-09 22:00:00 NaN
1994-10-10 22:00:00 NaN
1994-10-11 22:00:00 NaN
1994-10-12 22:00:00 NaN
1994-10-13 22:00:00 NaN
1994-10-14 22:00:00 NaN
1994-10-15 22:00:00 NaN
1994-10-16 22:00:00 NaN
1994-10-17 22:00:00 NaN
1994-10-18 22:00:00 NaN
1994-10-19 22:00:00 NaN
1994-10-20 22:00:00 NaN
1994-10-21 22:00:00 NaN
1994-10-22 22:00:00 NaN
1994-10-23 22:00:00 NaN
1994-10-24 22:00:00 NaN
1994-10-25 22:00:00 NaN
1994-10-26 22:00:00 NaN
1994-10-27 22:00:00 NaN
1994-10-28 22:00:00 NaN
1994-10-29 22:00:00 NaN
1994-10-30 22:00:00 1.29931
1994-10-31 22:00:00 1.29281
1994-11-01 22:00:00 1.27771
1994-11-02 22:00:00 1.27821
1994-11-03 22:00:00 1.28321
1994-11-04 22:00:00 NaN
1994-11-05 22:00:00 NaN
1994-11-06 22:00:00 1.28751
1994-11-07 22:00:00 1.27091
[225 rows x 1 columns]
원하는 출력하지만 이미 unchnaged 기간을 가지고 BID-CLOSE 값을 유지 NaN이 가득 어떤 날짜 간격을 가질 것이다. 아래 출력은 가상의 것이며 원하는 결과를 설명하기위한 것입니다.
BID-CLOSE
28/03/1994 22:00:00 1.15981
29/03/1994 22:00:00 1.16681
30/03/1994 22:00:00 1.15021
31/03/1994 22:00:00 1.14851
01/04/1994 21:00:00 NaN
02/04/1994 21:00:00 NaN
03/04/1994 21:00:00 1.13881
04/04/1994 21:00:00 1.14351
05/04/1994 21:00:00 1.14411
06/04/1994 21:00:00 1.14011
07/04/1994 21:00:00 1.14391
08/04/1994 21:00:00 NaN
09/04/1994 21:00:00 NaN
10/04/1994 21:00:00 1.14451
11/04/1994 21:00:00 1.14201
12/04/1994 21:00:00 1.13911
13/04/1994 21:00:00 1.14821
…
25/10/1994 21:00:00 1.29371
26/10/1994 21:00:00 NaN
27/10/1994 21:00:00 1.29681
28/10/1994 21:00:00 1.29971
29/10/1994 21:00:00 1.30411
30/10/1994 22:00:00 1.30311
31/10/1994 22:00:00 NaN
01/11/1994 22:00:00 NaN
02/11/1994 22:00:00 1.29341
코드가 미국 표준 시간대를 인식하도록하려면 어떻게해야합니까?
['date_range']에'tz '을 넘기는 경우에는 작동하지 않을 것입니다 (http://pandas.pydata.org/pandas-docs/stable/generated/pandas.date_range.html) ? 따라서 시간대가 일치하면 일치해야합니다. – EdChum