pandas 时间增量/时长

更新时间：2020-11-04 16:19:34 标签：pandas 时长

看过来

《pandas 教程》持续更新中，可作为 pandas 入门进阶课程、pandas 中文手册、用法大全，配有案例讲解和速查手册。提供建议、纠错、催更等加作者微信: gairuo123（备注：pandas教程）和关注公众号「盖若」ID: gairuo。查看更新日志。作者开办 Python 数据分析培训，详情 Python 数据分析培训。

本教程作者所著新书《深入浅出Pandas：利用Python进行数据处理与分析》（ISBN：9787111685456）已由机械工业出版社出版上市，各大电商平台有售，欢迎：查看详情并关注购买。

Timedelta 是时间差异，也就是时长，以差异单位表示，例如天，小时，分钟，秒。它们可以是正数的也可以是负数的。也就是两个时间的距离长短，这在业务中会经常出现，比如体育比赛中短跑的时长，可能是正数，也可能是负数，正数代表时间多多少，负数代表时间少多少。

生成时长数据

Timedelta 数据类型用来代表时间增量，两个固定时间相减会产生时差：

# 两个固定时间相减
pd.Timestamp('2020-11-01 15') - pd.Timestamp('2020-11-01 14')
# Timedelta('0 days 01:00:00')

pd.Timestamp('2020-11-01 08') - pd.Timestamp('2020-11-02 08')
# Timedelta('-1 days +00:00:00')

按以下格式传入字符串：

# 一天
pd.Timedelta('1 days')
# Timedelta('1 days 00:00:00')

pd.Timedelta('1 days 00:00:00')
# Timedelta('1 days 00:00:00')

pd.Timedelta('1 days 2 hours')
# Timedelta('1 days 02:00:00')

pd.Timedelta('-1 days 2 min 3us')
# Timedelta('-2 days +23:57:59.999997'

用关键字参数指定时间：

pd.Timedelta(days=5, seconds=10)
# Timedelta('5 days 00:00:10')

pd.Timedelta(minutes=3, seconds=2)
# Timedelta('0 days 00:03:02')

# 可以实现指定分钟有多少天，多少小时
pd.Timedelta(minutes=3242)

使用带周期量的偏移量别名：

# 一天
pd.Timedelta('1D')
# Timedelta('1 days 00:00:00')

# 两周
pd.Timedelta('2W')
# Timedelta('14 days 00:00:00')

# 一天2小时3分钟4秒
pd.Timedelta('1D2H3M4S')

带单位的整型数字：

# 一天
pd.Timedelta(1, unit='d')

# 100 秒
pd.Timedelta(100, unit='s')
# Timedelta('0 days 00:01:40')

# 4 周
pd.Timedelta(4, unit='w')
# Timedelta('28 days 00:00:00')

Python内置的datetime.timedelta或者Numpy的np.timedelta64：

import datetime
import numpy as np

# 一天10分钟
pd.Timedelta(datetime.timedelta(days=1, minutes=10))
# Timedelta('1 days 00:10:00')

# 100纳秒
pd.Timedelta(np.timedelta64(100, 'ns'))
# Timedelta('0 days 00:00:00.000000100')

负值：

# 负值
pd.Timedelta('-1min')
# Timedelta('-1 days +23:59:00')

# 空值，缺失值
pd.Timedelta('nan')
# NaT

# pd.Timedelta('nat')
# NaT

标准字符串（ISO 8601 Duration strings）：

# ISO 8601 Duration strings
pd.Timedelta('P0DT0H1M0S')
# Timedelta('0 days 00:01:00')

pd.Timedelta('P0DT0H0M0.000000123S')
# Timedelta('0 days 00:00:00.000000')

也可以用 DateOffsets (Day, Hour, Minute, Second, Milli, Micro, Nano) 来构建：

pd.Timedelta(pd.offsets.Second(2))
# Timedelta('0 days 00:00:02')

to_timedelta

可以直接生成单个时长数据：

pd.to_timedelta('1 days 06:05:01.00003')
# Timedelta('1 days 06:05:01.000030')
pd.to_timedelta('15.5us')
# Timedelta('0 days 00:00:00.000015')

pd.to_timedelta(pd.offsets.Day(3))
# Timedelta('3 days 00:00:00')

pd.to_timedelta('15.5min')
# Timedelta('0 days 00:15:30')

pd.to_timedelta(124524564574835)
# Timedelta('1 days 10:35:24.564574835')

类列表生成TimedeltaIndex数据：

pd.to_timedelta(['1 days 06:05:01.00003', '15.5us', 'nan'])
# TimedeltaIndex(['1 days 06:05:01.000030', '0 days 00:00:00.000015', NaT], dtype='timedelta64[ns]', freq=None)
pd.to_timedelta(np.arange(5), unit='s')
# TimedeltaIndex(['00:00:00', '00:00:01', '00:00:02', '00:00:03', '00:00:04'], dtype='timedelta64[ns]', freq=None)
pd.to_timedelta(np.arange(5), unit='d')
# TimedeltaIndex(['0 days', '1 days', '2 days', '3 days', '4 days'], dtype='timedelta64[ns]', freq=None)

如时间戳数据一样，时长数据的存储也有上下限：

pd.Timedelta.min
# Timedelta('-106752 days +00:12:43.145224')
pd.Timedelta.max
# Timedelta('106751 days 23:47:16.854775')

时长的操作

时长可以相加：

pd.Timedelta(pd.offsets.Day(2)) + pd.Timedelta(pd.offsets.Second(2)) +\
pd.Timedelta('00:00:00.000123')
# Timedelta('2 days 00:00:02.000123')

以下是一些操作示例：

s = pd.Series(pd.date_range('2012-1-1', periods=3, freq='D'))
td = pd.Series([pd.Timedelta(days=i) for i in range(3)])
df = pd.DataFrame({'A': s, 'B': td})
df
'''
           A      B
0 2012-01-01 0 days
1 2012-01-02 1 days
2 2012-01-03 2 days
'''
df['C'] = df['A'] + df['B']
df
'''
           A      B          C
0 2012-01-01 0 days 2012-01-01
1 2012-01-02 1 days 2012-01-03
2 2012-01-03 2 days 2012-01-05
'''
df.dtypes
'''
A     datetime64[ns]
B    timedelta64[ns]
C     datetime64[ns]
dtype: object
'''
s - s.max()
'''
0   -2 days
1   -1 days
2    0 days
dtype: timedelta64[ns]
'''
s - datetime.datetime(2011, 1, 1, 3, 5)
'''
0   364 days 20:55:00
1   365 days 20:55:00
2   366 days 20:55:00
dtype: timedelta64[ns]
'''
s + datetime.timedelta(minutes=5)
'''
0   2012-01-01 00:05:00
1   2012-01-02 00:05:00
2   2012-01-03 00:05:00
dtype: datetime64[ns]
'''
s + pd.offsets.Minute(5)
'''
0   2012-01-01 00:05:00
1   2012-01-02 00:05:00
2   2012-01-03 00:05:00
dtype: datetime64[ns]
'''
s + pd.offsets.Minute(5) + pd.offsets.Milli(5)
'''
0   2012-01-01 00:05:00.005
1   2012-01-02 00:05:00.005
2   2012-01-03 00:05:00.005
dtype: datetime64[ns]
'''

# 两个时间标量之间想减
y = s - s[0]
y
'''
0   0 days
1   1 days
2   2 days
dtype: timedelta64[ns]
'''
# 位移后
y = s - s.shift()
y
'''
0      NaT
1   1 days
2   1 days
dtype: timedelta64[ns]
'''
# 绝对值会将负数变为正数
td1 = pd.Timedelta('-1 days 2 hours 3 seconds')
abs(td1)
# Timedelta('1 days 02:00:03')

其他：

支持 min() 和 max()
支持 idxmin() 和 idxmax()
支持 y.fillna(pd.Timedelta(0))

时长序列的统计

时长序列可以做统计性计算：

y2 = pd.Series(pd.to_timedelta(['-1 days +00:00:05',
                                'nat',
                                '-1 days +00:00:05',
                                '1 days']))
y2
'''
0   -1 days +00:00:05
1                 NaT
2   -1 days +00:00:05
3     1 days 00:00:00
dtype: timedelta64[ns]
'''
y2.mean()
# Timedelta('-1 days +16:00:03.333333')
y2.median()
# Timedelta('-1 days +00:00:05')
y2.quantile(.1)
# Timedelta('-1 days +00:00:05')
y2.sum()
# Timedelta('-1 days +00:00:10')