如何使用 python 在时间序列数据中创建 3 个月的存储桶

Question

I have daily time-series data as date(YYYY-MM-dd) and quantity:

Date        Quantity
2017-10-31      5
2017-11-12      4
2017-11-13      7
2017-11-18      10
2017-12-03      28
2017-12-16      19
2018-01-03      24
2018-01-19      5
2018-02-02      1
2018-03-22      56
2018-04-12      12

I want to create 3 month buckets for date column. 3M column dates-names can be changed. In below table it represents 2017 between 10 and 12 months.

Date        Quantity    3M
2017-10-31      5       2017-10-12
2017-11-12      4       2017-10-12
2017-11-13      7       2017-10-12
2017-11-18      10      2017-10-12
2017-12-03      28      2017-10-12
2017-12-16      19      2017-10-12
2018-01-03      24      2018-01-03
2018-01-19      5       2018-01-03
2018-02-02      1       2018-01-03
2018-03-22      56      2018-01-03
2018-04-12      12      2018-04-06

How can I do that?

In the end I'll groupby 3M column with sum of Quantity column, so if there is direct way to do that that'll be great.

Any comment would be helpful Thanks,

Answer 1

As of Pandas v0.20.1 (May 5, 2017), pd.cut supports the datetime64 dtype.

from random import randrange
from datetime import timedelta, date
import numpy as np
import pandas as pd

def random_date(start, end):
    delta = end - start
    int_delta = (delta.days * 24 * 60 * 60) + delta.seconds
    random_second = randrange(int_delta)
    return start + timedelta(seconds=random_second)

start = date(2010, 1, 1)
end = date(2020, 1, 1)
df = pd.DataFrame({"dates": [np.datetime64(random_date(start, end)) for _ in range(20)]})
df
bins = pd.date_range(start, end, freq='3MS')

df["bins"] = pd.cut(df.dates, bins=bins)
df