第9章 时序数据
import pandas as pd
import numpy as np
一、时序的创建
1. 四类时间变量
现在理解可能关于③和④有些困惑,后面会作出一些说明
名称 |
描述 |
元素类型 |
创建方式 |
① Date times(时间点/时刻) |
描述特定日期或时间点 |
Timestamp |
to_datetime或date_range |
② Time spans(时间段/时期) |
由时间点定义的一段时期 |
Period |
Period或period_range |
③ Date offsets(相对时间差) |
一段时间的相对大小(与夏/冬令时无关) |
DateOffset |
DateOffset |
④ Time deltas(绝对时间差) |
一段时间的绝对大小(与夏/冬令时有关) |
Timedelta |
to_timedelta或timedelta_range |
2. 时间点的创建
(a)to_datetime方法
Pandas在时间点建立的输入格式规定上给了很大的自由度,下面的语句都能正确建立同一时间点
pd.to_datetime('2020.1.1')
pd.to_datetime('2020 1.1')
pd.to_datetime('2020 1 1')
pd.to_datetime('2020 1-1')
pd.to_datetime('2020-1 1')
pd.to_datetime('2020-1-1')
pd.to_datetime('2020/1/1')
pd.to_datetime('1.1.2020')
pd.to_datetime('1.1 2020')
pd.to_datetime('1 1 2020')
pd.to_datetime('1 1-2020')
pd.to_datetime('1-1 2020')
pd.to_datetime('1-1-2020')
pd.to_datetime('1/1/2020')
pd.to_datetime('20200101')
pd.to_datetime('2020.0101')
Timestamp('2020-01-01 00:00:00')
下面的语句都会报错
此时可利用format参数强制匹配
pd.to_datetime('2020\\1\\1',format='%Y\\%m\\%d')
pd.to_datetime('2020`1`1',format='%Y`%m`%d')
pd.to_datetime('2020.1 1',format='%Y.%m %d')
pd.to_datetime('1 1.2020',format='%d %m.%Y')
Timestamp('2020-01-01 00:00:00')
同时,使用列表可以将其转为时间点索引
pd.Series(range(2),index=pd.to_datetime(['2020/1/1','2020/1/2']))
2020-01-01 0
2020-01-02 1
dtype: int64
type(pd.to_datetime(['2020/1/1','2020/1/2']))
pandas.core.indexes.datetimes.DatetimeIndex
对于DataFrame而言,如果列已经按照时间顺序排好,则利用to_datetime可自动转换
df = pd.DataFrame({
'year': [2020, 2020],'month': [1, 1], 'day': [1, 2]})
pd.to_datetime(df)
0 2020-01-01
1 2020-01-02
dtype: datetime64[ns]
(b)时间精度与范围限制
事实上,Timestamp的精度远远不止day,可以最小到纳秒ns
pd.to_datetime('2020/1/1 00:00:00.123456789')
Timestamp('2020-01-01 00:00:00.123456789')
同时,它带来范围的代价就是只有大约584年的时间点是可用的
pd.Timestamp.min
Timestamp('1677-09-21 00:12:43.145225')
pd.Timestamp.max
Timestamp('2262-04-11 23:47:16.854775807')
(c)date_range方法
一般来说,start/end/periods(时间点个数)/freq(间隔方法)是该方法最重要的参数,给定了其中的3个,剩下的一个就会被确定
pd.date_range(start='2020/1/1',end='2020/1/10',periods=3)
DatetimeIndex(['2020-01-01 00:00:00', '2020-01-05 12:00:00',
'2020-01-10 00:00:00'],
dtype='datetime64[ns]', freq=None)
pd.date_range(start='2020/1/1',end='2020/1/10',freq='D')
DatetimeIndex(['2020-01-01', '2020-01-02', '2020-01-03', '2020-01-04',
'2020-01-05', '2020-01-06', '2020-01-07', '2020-01-08',
'2020-01-09', '2020-01-10'],
dtype='datetime64[ns]', freq='D')
pd.date_range(start='2020/1/1',periods=3,freq='D')
DatetimeIndex(['2020-01-01', '2020-01-02', '2020-01-03'], dtype='datetime64[ns]', freq='D')
pd.date_range(end='2020/1/3',periods=3,freq='D')
DatetimeIndex(['2020-01-01', '2020-01-02', '2020-01-03'], dtype='datetime64[ns]', freq='D')
其中freq参数有许多选项,下面将常用部分罗列如下,更多选项可看这里
符号 |
D/B |
W |
M/Q/Y |
BM/BQ/BY |
MS/QS/YS |
BMS/BQS/BYS |
H |
T |
S |
描述 |
日/工作日 |
周 |
月末 |
月/季/年末日 |
月/季/年末工作日 |
月/季/年初日 |
月/季/年初工作日 |
小时 |
分钟 |
pd.date_range(start='2020/1/1',periods=3,freq='T')
DatetimeIndex(['2020-01-01 00:00:00', '2020-01-01 00:01:00',
'2020-01-01 00:02:00'],
dtype='datetime64[ns]', freq='T')
pd.date_range(start='2020/1/1',periods=3,freq='M')
DatetimeIndex(['2020-01-31', '2020-02-29', '2020-03-31'], dtype='datetime64[ns]', freq='M')
pd.date_range(start='2020/1/1',periods=3,freq='BYS')
DatetimeIndex(['2020-01-01', '2021-01-01', '2022-01-03'], dtype='datetime64[ns]', freq='BAS-JAN')
bdate_range是一个类似与date_range的方法,特点在于可以在自带的工作日间隔设置上,再选择weekmask参数和holidays参数
它的freq中有一个特殊的’C’/‘CBM’/'CBMS’选项,表示定制,需要联合weekmask参数和holidays参数使用
例如现在需要将工作日中的周一、周二、周五3天保留,并将部分holidays剔除
weekmask = 'Mon Tue Fri'
holidays = [pd.Timestamp('2020/1/%s'%i) for i in range(7,13)]
pd.bdate_range(start='2020-1-1',end='2020-1-15',freq='C',weekmask=weekmask,holidays=holidays)
DatetimeIndex(['2020-01-03', '2020-01-06', '2020-01-13', '2020-01-14'], dtype='datetime64[ns]', freq='C')
3. DateOffset对象
(a)DataOffset与Timedelta的区别
Timedelta绝对时间差的特点指无论是冬令时还是夏令时,增减1day都只计算24小时
DataOffset相对时间差指,无论一天是23\24\25小时,增减1day都与当天相同的时间保持一致
例如,英国当地时间 2020年03月29日,01:00:00 时钟向前调整 1 小时 变为 2020年03月29日,02:00:00,开始夏令时
ts = pd.Timestamp('2020-3-29 01:00:00', tz='Europe/Helsinki')
ts + pd.Timedelta(days=1)
Timestamp('2020-03-30 02:00:00+0300', tz='Europe/Helsinki')
ts + pd.DateOffset(days=1)
Timestamp('2020-03-30 01:00:00+0300', tz='Europe/Helsinki')
这似乎有些令人头大,但只要把tz(time zone)去除就可以不用管它了,两者保持一致,除非要使用到时区变换
ts = pd.Timestamp('2020-3-29 01:00:00')
ts + pd.Timedelta(days=1)
Timestamp('2020-03-30 01:00:00')
ts + pd.DateOffset(days=1)
Timestamp('2020-03-30 01:00:00')
(b)增减一段时间
DateOffset的可选参数包括years/months/weeks/days/hours/minutes/seconds
pd.Timestamp('2020-01-01') + pd.DateOffset(minutes=20) - pd.DateOffset(weeks=2)
Timestamp('2019-12-18 00:20:00')
(c)各类常用offset对象
freq |
D/B |
W |
(B)M/(B)Q/(B)Y |
(B)MS/(B)QS/(B)YS |
H |
T |
S |
C |
|
offset |
DateOffset/BDay |
Week |
(B)MonthEnd/(B)QuarterEnd/(B)YearEnd |
(B)MonthBegin/(B)QuarterBegin/(B)YearBegin |
Hour |
Minute |
Second |
CDay(定制工作日) |
|
pd.Timestamp('2020-01-01') + pd.offsets.Week(2)
Timestamp('2020-01-15 00:00:00')
pd.Timestamp('2020-01-01') + pd.offsets.BQuarterBegin(1)
Timestamp('2020-03-02 00:00:00')
(d)序列的offset操作
利用apply函数
pd.Series(pd.offsets.BYearBegin(3).apply(i) for i in pd.date_range('20200101',periods=3,freq='Y'))
0 2023-01-02
1 2024-01-01
2 2025-01-01
dtype: datetime64[ns]
直接使用对象加减
pd.date_range('20200101',periods=3,freq='Y') + pd.offsets.BYearBegin(3)
DatetimeIndex(['2023-01-02', '2024-01-01', '2025-01-01'], dtype='datetime64[ns]', freq='A-DEC')
定制offset,可以指定weekmask和holidays参数(思考为什么三个都是一个值)
pd.Series(pd.offsets.CDay(3,weekmask='Wed Fri',holidays='2020010').apply(i)
for i in pd.date_range('20200105',periods=3,freq='D'))
0 2020-01-15
1 2020-01-15
2 2020-01-15
dtype: datetime64[ns]
二、时序的索引及属性
1. 索引切片
这一部分几乎与第二章的规则完全一致
rng = pd.date_range('2020','2021', freq='W')
ts = pd.Series(np.random.randn(len(rng)), index=rng)
ts.head()
2020-01-05 -0.275349
2020-01-12 2.359218
2020-01-19 -0.447633
2020-01-26 -0.479830
2020-02-02 0.517587
Freq: W-SUN, dtype: float64
ts['2020-01-26']
-0.47982974619679947
合法字符自动转换为时间点
ts['2020-01-26':'20200726'].head()
2020-01-26 -0.479830
2020-02-02 0.517587
2020-02-09 -0.575879
2020-02-16 0.952187
2020-02-23 0.554098
Freq: W-SUN, dtype: float64
2. 子集索引
ts['2020-7'].head()
2020-07-05 -0.088912
2020-07-12 0.153852
2020-07-19 1.670324
2020-07-26 0.568214
Freq: W-SUN, dtype: float64
支持混合形态索引
ts['2011-1':'20200726'].head()
2020-01-05 -0.275349
2020-01-12 2.359218
2020-01-19 -0.447633
2020-01-26 -0.479830
2020-02-02 0.517587
Freq: W-SUN, dtype: float64
3. 时间点的属性
采用dt对象可以轻松获得关于时间的信息
pd.Series(ts.index).dt.week.head()
0 1
1 2
2 3
3 4
4 5
dtype: int64
pd.Series(ts.index).dt.day.head()
0 5
1 12
2 19
3 26
4 2
dtype: int64
利用strftime可重新修改时间格式
pd.Series(ts.index).dt.strftime('%Y-间隔1-%m-间隔2-%d').head()
0 2020-间隔1-01-间隔2-05
1 2020-间隔1-01-间隔2-12
2 2020-间隔1-01-间隔2-19
3 2020-间隔1-01-间隔2-26
4 2020-间隔1-02-间隔2-02
dtype: object
对于datetime对象可以直接通过属性获取信息
pd.date_range('2020','2021', freq='W').month
Int64Index([ 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 3, 4, 4, 4, 4,
5, 5, 5, 5, 5, 6, 6, 6, 6, 7, 7, 7, 7, 8, 8, 8, 8,
8, 9, 9, 9, 9, 10, 10, 10, 10, 11, 11, 11, 11, 11, 12, 12, 12,
12],
dtype='int64')
pd.date_range('2020','2021', freq='W').weekday
Int64Index([6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,
6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,
6, 6, 6, 6, 6, 6, 6, 6],
dtype='int64')
三、重采样
所谓重采样,就是指resample函数,它可以看做时序版本的groupby函数
1. resample对象的基本操作
采样频率一般设置为上面提到的offset字符
df_r = pd.DataFrame(np.random.randn(1000, 3),index=pd.date_range('1/1/2020', freq='S', periods=1000),
columns=['A', 'B', 'C'])
r = df_r.resample('3min')
r
r.sum()
|
A |
B |
C |
2020-01-01 00:00:00 |
-8.772685 |
-27.074716 |
2.134617 |
2020-01-01 00:03:00 |
3.822484 |
8.912459 |
-15.448955 |
2020-01-01 00:06:00 |
2.744722 |
-8.055139 |
-11.364361 |
2020-01-01 00:09:00 |
4.655620 |
-11.524496 |
-10.536002 |
2020-01-01 00:12:00 |
-10.546811 |
5.063887 |
11.776490 |
2020-01-01 00:15:00 |
8.795150 |
-12.828809 |
-8.393950 |
df_r2 = pd.DataFrame(np.random.randn(200, 3),index=pd.date_range('1/1/2020', freq='D', periods=200),
columns=['A', 'B', 'C'])
r = df_r2.resample('CBMS')
r.sum()
|
A |
B |
C |
2020-01-01 |
5.278470 |
1.688588 |
5.904806 |
2020-02-03 |
-3.581797 |
7.515267 |
0.205308 |
2020-03-02 |
-5.021605 |
-4.441066 |
5.433917 |
2020-04-01 |
0.671702 |
3.840042 |
4.922487 |
2020-05-01 |
4.613352 |
9.702408 |
-4.928112 |
2020-06-01 |
-0.598191 |
7.387416 |
8.716921 |
2020-07-01 |
-0.327200 |
-1.577507 |
-3.956079 |
2. 采样聚合
r = df_r.resample('3T')
r['A'].mean()
2020-01-01 00:00:00 -0.048737
2020-01-01 00:03:00 0.021236
2020-01-01 00:06:00 0.015248
2020-01-01 00:09:00 0.025865
2020-01-01 00:12:00 -0.058593
2020-01-01 00:15:00 0.087952
Freq: 3T, Name: A, dtype: float64
r['A'].agg([np.sum, np.mean, np.std])
|
sum |
mean |
std |
2020-01-01 00:00:00 |
-8.772685 |
-0.048737 |
0.939954 |
2020-01-01 00:03:00 |
3.822484 |
0.021236 |
1.004048 |
2020-01-01 00:06:00 |
2.744722 |
0.015248 |
1.018865 |
2020-01-01 00:09:00 |
4.655620 |
0.025865 |
1.020881 |
2020-01-01 00:12:00 |
-10.546811 |
-0.058593 |
0.954328 |
2020-01-01 00:15:00 |
8.795150 |
0.087952 |
1.199379 |
类似地,可以使用函数/lambda表达式
r.agg({
'A': np.sum,'B': lambda x: max(x)-min(x)})
|
A |
B |
2020-01-01 00:00:00 |
-8.772685 |
4.950006 |
2020-01-01 00:03:00 |
3.822484 |
5.711679 |
2020-01-01 00:06:00 |
2.744722 |
6.923072 |
2020-01-01 00:09:00 |
4.655620 |
6.370589 |
2020-01-01 00:12:00 |
-10.546811 |
4.544878 |
2020-01-01 00:15:00 |
8.795150 |
5.244546 |
3. 采样组的迭代
采样组的迭代和groupby迭代完全类似,对于每一个组都可以分别做相应操作
small = pd.Series(range(6),index=pd.to_datetime(['2020-01-01 00:00:00', '2020-01-01 00:30:00'
, '2020-01-01 00:31:00','2020-01-01 01:00:00'
,'2020-01-01 03:00:00','2020-01-01 03:05:00']))
resampled = small.resample('H')
for name, group in resampled:
print("Group: ", name)
print("-" * 27)
print(group, end="\n\n")
Group: 2020-01-01 00:00:00
---------------------------
2020-01-01 00:00:00 0
2020-01-01 00:30:00 1
2020-01-01 00:31:00 2
dtype: int64
Group: 2020-01-01 01:00:00
---------------------------
2020-01-01 01:00:00 3
dtype: int64
Group: 2020-01-01 02:00:00
---------------------------
Series([], dtype: int64)
Group: 2020-01-01 03:00:00
---------------------------
2020-01-01 03:00:00 4
2020-01-01 03:05:00 5
dtype: int64
四、窗口函数
下面主要介绍pandas中两类主要的窗口(window)函数:rolling/expanding
s = pd.Series(np.random.randn(1000),index=pd.date_range('1/1/2020', periods=1000))
s.head()
2020-01-01 0.305974
2020-01-02 0.185221
2020-01-03 -0.646472
2020-01-04 -1.430293
2020-01-05 -0.956094
Freq: D, dtype: float64
1. Rolling
(a)常用聚合
所谓rolling方法,就是规定一个窗口,它和groupby对象一样,本身不会进行操作,需要配合聚合函数才能计算结果
s.rolling(window=50)
Rolling [window=50,center=False,axis=0]
s.rolling(window=50).mean()
2020-01-01 NaN
2020-01-02 NaN
2020-01-03 NaN
2020-01-04 NaN
2020-01-05 NaN
...
2022-09-22 0.160743
2022-09-23 0.136296
2022-09-24 0.147523
2022-09-25 0.133087
2022-09-26 0.130841
Freq: D, Length: 1000, dtype: float64
min_periods参数是指需要的非缺失数据点数量阀值
s.rolling(window=50,min_periods=3).mean().head()
2020-01-01 NaN
2020-01-02 NaN
2020-01-03 -0.051759
2020-01-04 -0.396392
2020-01-05 -0.508333
Freq: D, dtype: float64
count/sum/mean/median/min/max/std/var/skew/kurt/quantile/cov/corr都是常用的聚合函数
(b)rolling的apply聚合
使用apply聚合时,只需记住传入的是window大小的Series,输出的必须是标量即可,比如如下计算变异系数
s.rolling(window=50,min_periods=3).apply(lambda x:x.std()/x.mean()).head()
2020-01-01 NaN
2020-01-02 NaN
2020-01-03 -10.018809
2020-01-04 -2.040720
2020-01-05 -1.463460
Freq: D, dtype: float64
(c)基于时间的rolling
s.rolling('15D').mean().head()
2020-01-01 0.305974
2020-01-02 0.245598
2020-01-03 -0.051759
2020-01-04 -0.396392
2020-01-05 -0.508333
Freq: D, dtype: float64
可选closed=‘right’(默认)‘left’‘both’'neither’参数,决定端点的包含情况
s.rolling('15D', closed='right').sum().head()
2020-01-01 0.305974
2020-01-02 0.491195
2020-01-03 -0.155277
2020-01-04 -1.585570
2020-01-05 -2.541664
Freq: D, dtype: float64
2. Expanding
(a)expanding函数
普通的expanding函数等价与rolling(window=len(s),min_periods=1),是对序列的累计计算
s.rolling(window=len(s),min_periods=1).sum().head()
2020-01-01 0.305974
2020-01-02 0.491195
2020-01-03 -0.155277
2020-01-04 -1.585570
2020-01-05 -2.541664
Freq: D, dtype: float64
s.expanding().sum().head()
2020-01-01 0.305974
2020-01-02 0.491195
2020-01-03 -0.155277
2020-01-04 -1.585570
2020-01-05 -2.541664
Freq: D, dtype: float64
apply方法也是同样可用的
s.expanding().apply(lambda x:sum(x)).head()
2020-01-01 0.305974
2020-01-02 0.491195
2020-01-03 -0.155277
2020-01-04 -1.585570
2020-01-05 -2.541664
Freq: D, dtype: float64
(b)几个特别的Expanding类型函数
cumsum/cumprod/cummax/cummin都是特殊expanding累计计算方法
s.cumsum().head()
2020-01-01 0.305974
2020-01-02 0.491195
2020-01-03 -0.155277
2020-01-04 -1.585570
2020-01-05 -2.541664
Freq: D, dtype: float64
s.cumsum().head()
2020-01-01 0.305974
2020-01-02 0.491195
2020-01-03 -0.155277
2020-01-04 -1.585570
2020-01-05 -2.541664
Freq: D, dtype: float64
shift/diff/pct_change都是涉及到了元素关系
①shift是指序列索引不变,但值向后移动
②diff是指前后元素的差,period参数表示间隔,默认为1,并且可以为负
③pct_change是值前后元素的变化百分比,period参数与diff类似
s.shift(2).head()
2020-01-01 NaN
2020-01-02 NaN
2020-01-03 0.305974
2020-01-04 0.185221
2020-01-05 -0.646472
Freq: D, dtype: float64
s.diff(3).head()
2020-01-01 NaN
2020-01-02 NaN
2020-01-03 NaN
2020-01-04 -1.736267
2020-01-05 -1.141316
Freq: D, dtype: float64
s.pct_change(3).head()
2020-01-01 NaN
2020-01-02 NaN
2020-01-03 NaN
2020-01-04 -5.674559
2020-01-05 -6.161897
Freq: D, dtype: float64