这篇随笔主要介绍利用 Pandas 和 Datetime 处理时间序列。包括日期和时间数据类型的介绍、时间数据与字符串之间的转换、时间序列的索引、时间序列的平移、时期数据的创建和计算以及时间序列的重采样和频率转换
1
2
3
4
5
| import pandas as pd
from datetime import datetime
from datetime import timedelta
from dateutil.parser import parse
import numpy as np
|
日期和时间数据的类型
datetime 类型
1
2
| now = datetime.now()
now
|
1
| datetime.datetime(2021, 10, 9, 21, 38, 2, 642419)
|
1
| now.year, now.month, now.day, now.hour, now.minute, now.second
|
1
| (2021, 10, 9, 21, 38, 2)
|
timedelta 类型
1
2
| delta = datetime(2021, 5, 1, 0, 8, 40) - datetime(2019, 5, 1)
delta
|
1
| datetime.timedelta(days=731, seconds=520)
|
1
2
| start = datetime(2019, 5, 1)
start + timedelta(20, 20)
|
1
| datetime.datetime(2019, 5, 21, 0, 0, 20)
|
datetime 与 string 的转换
pd.to_datetime( list ) : datetime → string
to_datetime 方法可以解析多种不同的日期表示形式
1
2
| datestrs = ['2019/5/1/00:00:00', '2021-5-1 00:00:00']
pd.to_datetime(datestrs)
|
1
| DatetimeIndex(['2019-05-01', '2021-05-01'], dtype='datetime64[ns]', freq=None)
|
str(datetime), datetime.strftime('%Y-%m-%d') : datetime → string
1
2
| now = datetime.now()
now, str(now), now.strftime('%Y-%m-%d')
|
1
2
3
| (datetime.datetime(2021, 10, 9, 21, 38, 2, 880786),
'2021-10-09 21:38:02.880786',
'2021-10-09')
|
datetime 格式定义
| %Y, %y | 4 位数的年, 2 位数的年 |
| %m , %h | 2 位数的月 [01, 12], 英文缩写表示的月 [Jan, Feb, ..., Dec] |
| %d | 2 位数的日 [01, 31] |
| |
| %H , %l | 24 小时制的时 [00, 23], 12 小时制的时 [01, 12] |
| %M, %S | 2 位数的分 [00, 59], 2 位数的秒 [00, 59] |
| |
| %w | 整数表示的星期几 [0, 6] |
| %W | 2 位数的每年的第几周 [00， 53], 以星期一为每周的开始 |
| %U | 2 位数的每年的第几周 [00， 53], 以星期日为每周的开始 |
| |
| %F | %Y-%m-%d |
| %D | %m/%d/%y |
datetime.strptime(string, '%Y-%m-%d') : string → datetime
1
2
| value = '2019-05-01'
datetime.strptime(value, '%Y-%m-%d')
|
1
| datetime.datetime(2019, 5, 1, 0, 0)
|
1
2
| datestrs = ['7/25/1998', '11/24/1997']
[datetime.strptime(x, '%m/%d/%Y') for x in datestrs]
|
1
| [datetime.datetime(1998, 7, 25, 0, 0), datetime.datetime(1997, 11, 24, 0, 0)]
|
dateutil.parser.parse( string, dayfirst ) : string → datetime
1
| datetime.datetime(2019, 5, 1, 0, 0)
|
1
| parse('May 1, 2019 10:45 PM')
|
1
| datetime.datetime(2019, 5, 1, 22, 45)
|
在国际通用的格式中, 日出现在月的前面很普遍, 传入 dayfirst=True 即可解决这个问题
1
| parse('1/5/2019'), parse('1/5/2019', dayfirst=True)
|
1
| (datetime.datetime(2019, 1, 5, 0, 0), datetime.datetime(2019, 5, 1, 0, 0))
|
时间序列的索引
1
2
3
4
5
| dates = [datetime(2011, 1, 2), datetime(2011, 1, 5),
datetime(2011, 1, 7), datetime(2011, 1, 8),
datetime(2011, 1, 10), datetime(2011, 1, 12)]
ts = pd.Series(np.random.randn(6), index=dates)
ts
|
1
2
3
4
5
6
7
| 2011-01-02 -1.248703
2011-01-05 -0.072871
2011-01-07 1.672176
2011-01-08 -0.761810
2011-01-10 -0.044322
2011-01-12 -0.701274
dtype: float64
|
可以传入一个可以被解释为日期的字符串进行切片
1
2
3
| ts = pd.Series(np.arange(1000),
index=pd.date_range('1/1/2000', periods=1000))
ts.head()
|
1
2
3
4
5
6
| 2000-01-01 0
2000-01-02 1
2000-01-03 2
2000-01-04 3
2000-01-05 4
Freq: D, dtype: int32
|
1
| ts['2000-1-1':'2000-1-3']
|
1
2
3
4
| 2000-01-01 0
2000-01-02 1
2000-01-03 2
Freq: D, dtype: int32
|
对于较长的时间序列, 只需传入“年”或“年月”即可轻松选取数据的切片
1
2
3
4
5
6
| 2001-01-01 366
2001-01-02 367
2001-01-03 368
2001-01-04 369
2001-01-05 370
Freq: D, dtype: int32
|
1
2
3
4
5
6
| 2001-02-01 397
2001-02-02 398
2001-02-03 399
2001-02-04 400
2001-02-05 401
Freq: D, dtype: int32
|
传入 datetime 对象也可以进行切片
1
| ts[datetime(2001,1,7):datetime(2001,1,8)]
|
1
2
3
| 2001-01-07 372
2001-01-08 373
Freq: D, dtype: int32
|
ts.truncate( before, after )
before : 向后切片
1
| ts.truncate(before='2001-9').head()
|
1
2
3
4
5
6
| 2001-09-01 609
2001-09-02 610
2001-09-03 611
2001-09-04 612
2001-09-05 613
Freq: D, dtype: int32
|
after : 向前切片
1
| ts.truncate(after='2000-2-4').head()
|
1
2
3
4
5
6
| 2000-01-01 0
2000-01-02 1
2000-01-03 2
2000-01-04 3
2000-01-05 4
Freq: D, dtype: int32
|
1
| ts.truncate(before = '2000-1-1', after='2000-1-4')
|
1
2
3
4
5
| 2000-01-01 0
2000-01-02 1
2000-01-03 2
2000-01-04 3
Freq: D, dtype: int32
|
时间的范围、频率和位移
pd.date_range( start, end, periods, freq )
1
| pd.date_range('2019-05-01', '2021-05-01')
|
1
2
3
4
5
6
7
8
| DatetimeIndex(['2019-05-01', '2019-05-02', '2019-05-03', '2019-05-04',
'2019-05-05', '2019-05-06', '2019-05-07', '2019-05-08',
'2019-05-09', '2019-05-10',
...
'2021-04-22', '2021-04-23', '2021-04-24', '2021-04-25',
'2021-04-26', '2021-04-27', '2021-04-28', '2021-04-29',
'2021-04-30', '2021-05-01'],
dtype='datetime64[ns]', length=732, freq='D')
|
1
| pd.date_range('2019-05-01', periods=3)
|
1
| DatetimeIndex(['2019-05-01', '2019-05-02', '2019-05-03'], dtype='datetime64[ns]', freq='D')
|
freq = 'BM' : 'business end of month' 每月最后一个工作日
| H, T/min, S, L/ms, U | 每小时, 每分钟, 每秒, 每毫秒, 每微秒 |
| D, M, MS | 每日历日, 每月的最后一个日历日, 每月的第一个日历日 |
| B, BM, BMS | 每工作日, 每月的最后一个工作日, 每月的第一个工作日 |
| W-MON, W-TUE, ..., W-SUN | 每周, 从指定的星期几开始算起 |
| WOM-1MON, WOM-2MON, ..., WOM-4SUN | 每月, 从指定的第几周的星期几开始算起 |
| Q-JAN, Q-FEB, ..., Q-DEC | 每季, 从指定的月份的最后一个日历日开始算起 |
| QS-JAN, QS-FEB, ..., QS-DEC | 每季, 从指定的月份的第一个日历日开始算起 |
| BQ-JAN, BQ-FEB, ..., BQ-DEC | 每季, 从指定的月份的最后一个工作日开始算起 |
| BQS-JAN, BQS-FEB, ..., BQS-DEC | 每季, 从指定的月份的第一个工作日开始算起 |
| A-JAN, A-FEB, ..., A-DEC | 每年, 从指定的月份的最后一个日历日开始算起 |
| AS-JAN, AS-FEB, ..., AS-DEC | 每年, 从指定的月份的第一个日历日开始算起 |
| BA-JAN, BA-FEB, ..., BA-DEC | 每年, 从指定的月份的最后一个工作日开始算起 |
| BAS-JAN, BAS-FEB, ..., BAS-DEC | 每年, 从指定的月份的第一个工作日开始算起 |
1
| pd.date_range('2020-01-01', '2021-01-01', freq='M')
|
1
2
3
4
| DatetimeIndex(['2020-01-31', '2020-02-29', '2020-03-31', '2020-04-30',
'2020-05-31', '2020-06-30', '2020-07-31', '2020-08-31',
'2020-09-30', '2020-10-31', '2020-11-30', '2020-12-31'],
dtype='datetime64[ns]', freq='M')
|
1
| pd.date_range('2020-01-01', '2021-01-01', freq='4M')
|
DatetimeIndex(['2020-01-31', '2020-05-31', '2020-09-30'], dtype='datetime64[ns]', freq='4M')
1
| pd.date_range('2020-01-01', '2020-01-02', freq='2h30min')
|
1
2
3
4
5
6
| DatetimeIndex(['2020-01-01 00:00:00', '2020-01-01 02:30:00',
'2020-01-01 05:00:00', '2020-01-01 07:30:00',
'2020-01-01 10:00:00', '2020-01-01 12:30:00',
'2020-01-01 15:00:00', '2020-01-01 17:30:00',
'2020-01-01 20:00:00', '2020-01-01 22:30:00'],
dtype='datetime64[ns]', freq='150T')
|
ts.shift( periods, freq ) : 时间序列的平移
1
2
3
| ts = pd.Series(np.arange(1, 5),
index=pd.date_range('1/1/2000', periods=4, freq='D'))
ts
|
1
2
3
4
5
| 2000-01-01 1
2000-01-02 2
2000-01-03 3
2000-01-04 4
Freq: D, dtype: int32
|
Series 和 DataFrame 都有一个 shift 方法用于执行单纯的前移或后移操作, 保持索引不变
1
| ts.shift(-1), ts / ts.shift(1) -1
|
1
2
3
4
5
6
7
8
9
10
| (2000-01-01 2.0
2000-01-02 3.0
2000-01-03 4.0
2000-01-04 NaN
Freq: D, dtype: float64,
2000-01-01 NaN
2000-01-02 1.000000
2000-01-03 0.500000
2000-01-04 0.333333
Freq: D, dtype: float64)
|
ts.shift(-1) vs ts.shfit(-1, freq='D')
1
2
3
4
5
| 1999-12-31 1
2000-01-01 2
2000-01-02 3
2000-01-03 4
Freq: D, dtype: int32
|
1
2
3
4
5
| 2000-01-04 1
2000-01-05 2
2000-01-06 3
2000-01-07 4
Freq: D, dtype: int32
|
1
2
3
4
5
| 2000-01-01 01:30:00 1
2000-01-02 01:30:00 2
2000-01-03 01:30:00 3
2000-01-04 01:30:00 4
Freq: D, dtype: int32
|
Hour(), Minute(), Day(), MonthEnd() : 通过偏移量对日期进行平移
Hour(), Minute(), Day(), MonthEnd(), ... : 日期偏移量（ date offset ）对象
1
2
3
| from pandas.tseries.offsets import Day, MonthEnd
now = datetime.now()
now
|
1
| datetime.datetime(2021, 10, 9, 21, 38, 3, 837907)
|
1
| now + 3 * Day(), now + Day(3)
|
1
2
| (Timestamp('2021-10-12 21:38:03.837907'),
Timestamp('2021-10-12 21:38:03.837907'))
|
1
| Timestamp('2021-10-31 21:38:03.837907')
|
offset.rollforward( datetime ), offset.rollback( datetime ) : 控制将日期向前或向后“滚动”
1
2
| offset = MonthEnd()
offset.rollforward(now), offset.rollback(now)
|
1
2
| (Timestamp('2021-10-31 21:38:03.837907'),
Timestamp('2021-09-30 21:38:03.837907'))
|
1
2
| ts = pd.Series(np.ones(20), index=pd.date_range('1/15/2000', periods=20, freq='4d'))
ts
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
| 2000-01-15 1.0
2000-01-19 1.0
2000-01-23 1.0
2000-01-27 1.0
2000-01-31 1.0
2000-02-04 1.0
2000-02-08 1.0
2000-02-12 1.0
2000-02-16 1.0
2000-02-20 1.0
2000-02-24 1.0
2000-02-28 1.0
2000-03-03 1.0
2000-03-07 1.0
2000-03-11 1.0
2000-03-15 1.0
2000-03-19 1.0
2000-03-23 1.0
2000-03-27 1.0
2000-03-31 1.0
Freq: 4D, dtype: float64
|
ts.groupby( offset.rollforward )
1
2
| offset = MonthEnd()
ts.groupby(MonthEnd().rollforward).sum()
|
1
2
3
4
| 2000-01-31 5.0
2000-02-29 7.0
2000-03-31 8.0
dtype: float64
|
ts.resample( freq )
1
2
3
4
| 2000-01-31 1.0
2000-02-29 1.0
2000-03-31 1.0
Freq: M, dtype: float64
|
时期 (period) 及其计算
pd.Period( value, freq ) : 时期的创建
1
2
| p = pd.Period(2021, freq='A-MAY')
p
|
1
| p - pd.Period(2019, freq='A-MAY')
|
pd.period_range( start, end, freq ) : 时期范围的创建
1
2
| rng = pd.period_range('2000-01-01', '2000-06-30', freq='M')
rng
|
1
| PeriodIndex(['2000-01', '2000-02', '2000-03', '2000-04', '2000-05', '2000-06'], dtype='period[M]')
|
1
| pd.Series(np.arange(6), index=rng)
|
1
2
3
4
5
6
7
| 2000-01 0
2000-02 1
2000-03 2
2000-04 3
2000-05 4
2000-06 5
Freq: M, dtype: int32
|
pd.PeriodIndex( values, freq, year, quarter ) : 季度时期范围的创建
1
2
| values = ['2001Q3', '2002Q2', '2003Q1']
pd.PeriodIndex(values, freq='Q-DEC')
|
1
| PeriodIndex(['2001Q3', '2002Q2', '2003Q1'], dtype='period[Q-DEC]')
|
通过数组创建 PeriodIndex
1
2
| data = pd.read_csv('pydata-book-2nd-edition/examples/macrodata.csv')
data.year, data.quarter
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
| (0 1959.0
1 1959.0
2 1959.0
3 1959.0
4 1960.0
...
198 2008.0
199 2008.0
200 2009.0
201 2009.0
202 2009.0
Name: year, Length: 203, dtype: float64,
0 1.0
1 2.0
2 3.0
3 4.0
4 1.0
...
198 3.0
199 4.0
200 1.0
201 2.0
202 3.0
Name: quarter, Length: 203, dtype: float64)
|
1
2
3
| index = pd.PeriodIndex(year=data.year, quarter=data.quarter,
freq='Q-DEC')
index
|
1
2
3
4
5
6
| PeriodIndex(['1959Q1', '1959Q2', '1959Q3', '1959Q4', '1960Q1', '1960Q2',
'1960Q3', '1960Q4', '1961Q1', '1961Q2',
...
'2007Q2', '2007Q3', '2007Q4', '2008Q1', '2008Q2', '2008Q3',
'2008Q4', '2009Q1', '2009Q2', '2009Q3'],
dtype='period[Q-DEC]', length=203)
|
| year | quarter | realgdp | realcons | realinv | realgovt | realdpi | cpi | m1 | tbilrate | unemp | pop | infl | realint |
|---|
| 1959Q1 | 1959.0 | 1.0 | 2710.349 | 1707.4 | 286.898 | 470.045 | 1886.9 | 28.980 | 139.7 | 2.82 | 5.8 | 177.146 | 0.00 | 0.00 |
|---|
| 1959Q2 | 1959.0 | 2.0 | 2778.801 | 1733.7 | 310.859 | 481.301 | 1919.7 | 29.150 | 141.7 | 3.08 | 5.1 | 177.830 | 2.34 | 0.74 |
|---|
| 1959Q3 | 1959.0 | 3.0 | 2775.488 | 1751.8 | 289.226 | 491.260 | 1916.4 | 29.350 | 140.5 | 3.82 | 5.3 | 178.657 | 2.74 | 1.09 |
|---|
| 1959Q4 | 1959.0 | 4.0 | 2785.204 | 1753.7 | 299.356 | 484.052 | 1931.3 | 29.370 | 140.0 | 4.33 | 5.6 | 179.386 | 0.27 | 4.06 |
|---|
| 1960Q1 | 1960.0 | 1.0 | 2847.699 | 1770.5 | 331.722 | 462.199 | 1955.5 | 29.540 | 139.6 | 3.50 | 5.2 | 180.007 | 2.31 | 1.19 |
|---|
| ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... |
|---|
| 2008Q3 | 2008.0 | 3.0 | 13324.600 | 9267.7 | 1990.693 | 991.551 | 9838.3 | 216.889 | 1474.7 | 1.17 | 6.0 | 305.270 | -3.16 | 4.33 |
|---|
| 2008Q4 | 2008.0 | 4.0 | 13141.920 | 9195.3 | 1857.661 | 1007.273 | 9920.4 | 212.174 | 1576.5 | 0.12 | 6.9 | 305.952 | -8.79 | 8.91 |
|---|
| 2009Q1 | 2009.0 | 1.0 | 12925.410 | 9209.2 | 1558.494 | 996.287 | 9926.4 | 212.671 | 1592.8 | 0.22 | 8.1 | 306.547 | 0.94 | -0.71 |
|---|
| 2009Q2 | 2009.0 | 2.0 | 12901.504 | 9189.0 | 1456.678 | 1023.528 | 10077.5 | 214.469 | 1653.6 | 0.18 | 9.2 | 307.226 | 3.37 | -3.19 |
|---|
| 2009Q3 | 2009.0 | 3.0 | 12990.341 | 9256.0 | 1486.398 | 1044.088 | 10040.6 | 216.385 | 1673.9 | 0.12 | 9.6 | 308.013 | 3.56 | -3.44 |
|---|
203 rows × 14 columns
period.asfreq( freq, how ) : 时期的频率转换
年 → 月
1
2
| p = pd.Period('2021', freq='A-MAY')
p, p.asfreq('M', how='start'), p.asfreq('M', how='end')
|
1
| (Period('2021', 'A-MAY'), Period('2020-06', 'M'), Period('2021-05', 'M'))
|
月 → 年
1
2
| p = pd.Period('May-2021', 'M')
p, p.asfreq('A-DEC')
|
1
| (Period('2021-05', 'M'), Period('2021', 'A-DEC'))
|
年 → 日
1
2
3
| rng = pd.period_range('2006', '2009', freq='A-DEC')
ts = pd.Series(np.random.randn(len(rng)), index=rng)
ts, ts.asfreq('M', how='start'), ts.asfreq('D', how='start')
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
| (2006 -0.954789
2007 2.045184
2008 -0.367045
2009 -0.672529
Freq: A-DEC, dtype: float64,
2006-01 -0.954789
2007-01 2.045184
2008-01 -0.367045
2009-01 -0.672529
Freq: M, dtype: float64,
2006-01-01 -0.954789
2007-01-01 2.045184
2008-01-01 -0.367045
2009-01-01 -0.672529
Freq: D, dtype: float64)
|
按季度计算的时期
1
2
| p = pd.Period('2021Q1', freq='Q-DEC')
p, p.asfreq('D', 'start'), p.asfreq('D', 'end')
|
1
2
3
| (Period('2021Q1', 'Q-DEC'),
Period('2021-01-01', 'D'),
Period('2021-03-31', 'D'))
|
1
2
| p = pd.Period('2021Q1', freq='Q-JUN')
p, p.asfreq('D', 'start'), p.asfreq('D', 'end')
|
1
2
3
| (Period('2021Q1', 'Q-JUN'),
Period('2020-07-01', 'D'),
Period('2020-09-30', 'D'))
|
1
2
| p = pd.Period('2021Q4', freq='Q-JUN')
p, p.asfreq('D', 'start'), p.asfreq('D', 'end')
|
1
2
3
| (Period('2021Q4', 'Q-JUN'),
Period('2021-04-01', 'D'),
Period('2021-06-30', 'D'))
|
获取该季度倒数第二个工作日下午4点的时间戳 :
1
2
3
| from pandas.tseries.offsets import Hour
p4pm = (p.asfreq('B', 'end') - 1).asfreq('T', 'start') + Hour(16)
p4pm, p4pm.to_timestamp()
|
1
| (Period('2021-06-29 16:00', 'T'), Timestamp('2021-06-29 16:00:00'))
|
季度型范围的算术运算
1
2
3
| rng = pd.period_range('2019Q3', '2021Q4', freq='Q-DEC')
ts = pd.Series(np.arange(len(rng)), index=rng)
ts
|
1
2
3
4
5
6
7
8
9
10
11
| 2019Q3 0
2019Q4 1
2020Q1 2
2020Q2 3
2020Q3 4
2020Q4 5
2021Q1 6
2021Q2 7
2021Q3 8
2021Q4 9
Freq: Q-DEC, dtype: int32
|
1
2
3
| new_rng = rng.asfreq('D', 'end').asfreq('T', 'start') + 16 * 60
ts.index = new_rng.to_timestamp()
ts
|
1
2
3
4
5
6
7
8
9
10
11
| 2019-09-30 16:00:00 0
2019-12-31 16:00:00 1
2020-03-31 16:00:00 2
2020-06-30 16:00:00 3
2020-09-30 16:00:00 4
2020-12-31 16:00:00 5
2021-03-31 16:00:00 6
2021-06-30 16:00:00 7
2021-09-30 16:00:00 8
2021-12-31 16:00:00 9
Freq: Q-DEC, dtype: int32
|
ts.to_period( freq, copy ) : 时间戳 → 时期
1
2
3
| rng = pd.date_range('2021-01-01', periods=3, freq='M')
ts = pd.Series(np.random.randn(3), index=rng)
ts, ts.to_period()
|
1
2
3
4
5
6
7
8
| (2021-01-31 0.799601
2021-02-28 -0.747543
2021-03-31 -0.333042
Freq: M, dtype: float64,
2021-01 0.799601
2021-02 -0.747543
2021-03 -0.333042
Freq: M, dtype: float64)
|
1
2
3
| rng = pd.date_range('1/29/2021', periods=6, freq='D')
ts2 = pd.Series(np.random.randn(6), index=rng)
ts2, ts2.to_period('M')
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
| (2021-01-29 -0.917420
2021-01-30 -1.667312
2021-01-31 0.370422
2021-02-01 -1.841920
2021-02-02 0.067595
2021-02-03 0.539212
Freq: D, dtype: float64,
2021-01 -0.917420
2021-01 -1.667312
2021-01 0.370422
2021-02 -1.841920
2021-02 0.067595
2021-02 0.539212
Freq: M, dtype: float64)
|
ts.to_timestamp( freq, how, copy ) : 时期 → 时间戳
1
2
3
| rng = pd.period_range('1/29/2021', periods=2, freq='M')
ts3 = pd.Series(np.random.randn(2), index=rng)
ts3, ts3.to_timestamp()
|
1
2
3
4
5
6
| (2021-01 -2.029517
2021-02 -0.154121
Freq: M, dtype: float64,
2021-01-01 -2.029517
2021-02-01 -0.154121
dtype: float64)
|
重采样和频率转换
ts.resample( freq, axis, closed, label, kind, fill_method, limit, convention )
1
2
3
| rng = pd.date_range('2000-01-01', periods=100, freq='D')
ts = pd.Series(np.ones(len(rng)), index=rng)
ts
|
1
2
3
4
5
6
7
8
9
10
11
12
| 2000-01-01 1.0
2000-01-02 1.0
2000-01-03 1.0
2000-01-04 1.0
2000-01-05 1.0
...
2000-04-05 1.0
2000-04-06 1.0
2000-04-07 1.0
2000-04-08 1.0
2000-04-09 1.0
Freq: D, Length: 100, dtype: float64
|
1
2
3
4
5
| 2000-01-31 31.0
2000-02-29 29.0
2000-03-31 31.0
2000-04-30 9.0
Freq: M, dtype: float64
|
kind : 聚合到周期 ( 'period' ) 或 时间戳 ( 'timestamp' )
1
| ts.resample('M', kind='period').sum()
|
1
2
3
4
5
| 2000-01 31.0
2000-02 29.0
2000-03 31.0
2000-04 9.0
Freq: M, dtype: float64
|
closed : 降采样中, 设置时间闭合的一端, 'right' (start, end] 或 'left' [start, end)
1
2
3
| rng2 = pd.date_range('2000-01-01', periods=7, freq='T')
ts2 = pd.Series(np.arange(7), index=rng2)
ts2
|
1
2
3
4
5
6
7
8
| 2000-01-01 00:00:00 0
2000-01-01 00:01:00 1
2000-01-01 00:02:00 2
2000-01-01 00:03:00 3
2000-01-01 00:04:00 4
2000-01-01 00:05:00 5
2000-01-01 00:06:00 6
Freq: T, dtype: int32
|
1
| ts2.resample('5T', closed='right').sum()
|
1
2
3
4
| 1999-12-31 23:55:00 0
2000-01-01 00:00:00 15
2000-01-01 00:05:00 6
Freq: 5T, dtype: int32
|
1
| ts2.resample('5T', closed='left').sum()
|
1
2
3
| 2000-01-01 00:00:00 10
2000-01-01 00:05:00 11
Freq: 5T, dtype: int32
|
label : 降采样中, 设置聚合值的标签, 'right' 或 'left'
1
| ts2.resample('5T', closed='right', label='right').sum()
|
1
2
3
4
| 2000-01-01 00:00:00 0
2000-01-01 00:05:00 15
2000-01-01 00:10:00 6
Freq: 5T, dtype: int32
|
convention : 升采样中, 设置低频周期对应的高频时间戳, 'start' 或 'end'
1
2
3
4
5
| frame = pd.DataFrame(np.random.randn(2, 4),
index=pd.date_range('1/1/2000', periods=2,
freq='M'),
columns=['Colorado', 'Texas', 'New York', 'Ohio'])
frame
|
| Colorado | Texas | New York | Ohio |
|---|
| 2000-01-31 | 1.025930 | -0.974227 | 0.842722 | 1.017311 |
|---|
| 2000-02-29 | 1.171868 | -1.525096 | -0.159692 | 0.148648 |
|---|
1
2
| df_month = frame.resample('M', kind='period').sum()
df_month
|
| Colorado | Texas | New York | Ohio |
|---|
| 2000-01 | 1.025930 | -0.974227 | 0.842722 | 1.017311 |
|---|
| 2000-02 | 1.171868 | -1.525096 | -0.159692 | 0.148648 |
|---|
1
2
| df_daily = df_month.resample('D', convention='start').asfreq()
df_daily.head()
|
| Colorado | Texas | New York | Ohio |
|---|
| 2000-01-01 | 1.02593 | -0.974227 | 0.842722 | 1.017311 |
|---|
| 2000-01-02 | NaN | NaN | NaN | NaN |
|---|
| 2000-01-03 | NaN | NaN | NaN | NaN |
|---|
| 2000-01-04 | NaN | NaN | NaN | NaN |
|---|
| 2000-01-05 | NaN | NaN | NaN | NaN |
|---|