It is converted into a time series having a fixed frequency of time series
Just call resample to:
import datetime
import pandas as pd
import numpy as np
dates = [
datetime.datetime(2011, 1, 2),
datetime.datetime(2011, 1, 5),
datetime.datetime(2011, 1, 7),
datetime.datetime(2011, 1, 8),
datetime.datetime(2011, 1, 10),
datetime.datetime(2011, 1, 12),
]
ts = pd.Series(np.random.randn(6), index=dates)
dts = ts.resample("D")
# print(dts) # DatetimeIndexResampler [freq=<Day>, axis=0, closed=left, label=left, convention=start, base=0]
Use basic frequency
Generation time range:
# 默认情况下 传入起止时间会生成按天计算的时间点
index = pd.date_range('4/1/2012', '6/1/2012')
# print(index
# 如果我们只传入开始或者是结束时间 就要传入一个表示一段时间的数字
idx1 = pd.date_range(start='4/1/2012', periods=20)
# print(idx1)
idx2 = pd.date_range(end='6/1/2012', periods=20)
# print(idx2)
Start time and end time of strict boundaries defined time index: BM (business end of month)
so that only contain (or just point at the border) within the time interval
ret1 = pd.date_range('1/1/2000', '12/1/2000', freq="BM")
# print(ret1)
# date_range默认会保留起始和结束时间戳的时间信息
ret2 = pd.date_range('5/2/2012 12:56:31', periods=5)
# print(ret2)
# 这时候如果我们想要的是一组被规范到午夜的时间戳 normalize 可以实现
ret3 = pd.date_range('5/2/2012 12:56:31', periods=5, normalize=True)
# print(ret3)
Date and frequency offset
Explicitly create an object,
hour = pd.tseries.offsets.Hour()
# print(hour)
four_hour = pd.tseries.offsets.Hour(4)
# print(four_hour)
# 大部分的偏移对象可以通过加法进行连接
res = pd.tseries.offsets.Hour(4) + pd.tseries.offsets.Minute(30)
# print(res)
# 一般来说我们并不需要显式创建这样的对象
dts1 = pd.date_range('1/1/2000', '1/3/2000 23:59', freq="4h")
# print(dts1)
dts2 = pd.date_range('1/1/2000', periods=10, freq="2h30min")
# print(dts2)
Offset anchor
For example, say "M" is at the end of the calendar "BM" is the last working day of each month and some time frequency described are not evenly distributed
WOM date
WOM (week of month) is the frequency of use of a very class that allows us to get dates, such as "every third Friday" like
the following which is the standard monthly expiration date of stock options
rng = pd.date_range('1/1/2012', '9/1/2012', freq="WOM-3FRI")
# print(rng)
# print(list(rng))
Mobile (leading and lagging) data
Refers to the movement of advance along the time axis or the shift data
shifted index remains unchanged after a series shift and df has a simple method for performing a forward or
ts = pd.Series(np.random.randn(4), index=pd.date_range('1/1/2000', periods=4, freq="M"))
# print(ts)
# 向下平移
ts1 = ts.shift(2)
# print(ts1)
# 向上平移
ts2 = ts.shift(-2)
# print(ts2)
# shift 通常是用来计算一个或者多个时间序列的百分比变化
ret = ts/ts.shift(1) - 1
# print(ret)
"""
单纯的位移不会修改索引 所以部分数据会被丢弃
如果频率已经知道 可以经其传递给 shift 以便实现对时间的位移而不是对数据进行简单位移
(只是移动时间轴 不移动数据）
"""
# print(ts)
# print(ts.shift(2, freq="M"))
# print(ts.shift(3, freq="D"))
# print(ts.shift(1, freq="3D"))
# print(ts.shift(1, freq="90T"))
Date displaced by an offset of
from pandas.tseries.offsets import Day, MonthEnd, QuarterBegin, QuarterEnd
now = datetime.datetime(2011, 11, 17)
t1 = now + 3 * Day()
# print(t1)
# 等价于 datetime 中的 timedelta
t2 = now + datetime.timedelta(days=3)
# print(t2)
# print(t1 == t2)
# 如果加的是锚点偏移量 第一次增量会将原来的日期向前滚动到符合频率的规则的下一个日期
dt1 = now + MonthEnd()
# print(dt1)
dt2 = now - QuarterEnd()
# print(dt2)
# 也可以这样用 使用 锚点偏移量的 rollforward 和 rollback 方法
offset = MonthEnd()
tt1 = offset.rollback(now)
# print(tt1)
tt2 = offset.rollforward(now)
# print(tt2)
# 结合 groupby 使用日期偏移量
ts = pd.Series(np.random.randn(20), index=pd.date_range("1/15/2000", periods=20, freq='4d'))
# print(ts)
# 前向推出所有的月末数据 然后再分组求平均
ret2 = ts.groupby(offset.rollforward).mean()
# print(ret2)
# 更简单的方法是使用 resample 然后再使用 mean
ret3 = ts.resample("M", how="mean")
print(ret3)