python - 我试图将公司交易量的平均值与同一家公司的每日交易量进行比较，并找出 Pandas 的差异。我在 company 上做了 groupby

Question

我试图将公司交易量的平均值与同一家公司的每日交易量进行比较，并找出 Pandas 的差异。我已经对公司进行了分组，并获得了每个公司数量的平均值。我希望将平均值与同一家公司的每日交易量进行比较。
下面的代码是:

vol_grp.mean()

输出是:

            Volume
Company 
20MICRONS   947802.086957
3MINDIA     3881.608696
5PAISA      69606.521739
AAKASH      49254.217391
AARON       46435.583333
... ...
ZODJRDMKJ   50541.666667
ZOTA        36271.130435
ZUARI       285558.652174
ZUARIGLOB   149646.347826
ZYDUSWELL   72017.826087
1397 rows × 1 columns

实际数据是

    Date           Company      Volume
1   03-MAY-2021   20MICRONS     192281
4   03-MAY-2021   3MINDIA       1707
7   03-MAY-2021   5PAISA        81581
16  03-MAY-2021   AAKASH        35865
17  03-MAY-2021   AARON         1255
... ... ... ...
47160   03-JUN-2021 ZODIACLOTH  75966
47162   03-JUN-2021 ZOTA        470978
47163   03-JUN-2021 ZUARI       137563
47164   03-JUN-2021 ZUARIGLOB   51545
47165   03-JUN-2021 ZYDUSWELL   24350

例如，我有 20MICRONS 公司的意思，我想与 20MICRONS 公司的每日交易量进行比较。如果我有 20MICRONS 体积的 30 天信息，我的平均值应该与这 30 个值进行比较，并且应该返回 30 个差异值。对于所有其他公司也是如此

Answer 1

所以我创建了这个示例 df

from datetime import datetime as dt
import pandas as pd
from numpy.random import randint
df = pd.DataFrame(dict(date= [dt(2021,1,i) for i in [1]*4+[2]*4+[3]*4], 
                       company= ["AAPL", "FB", "NVDA", "AMZN"]*3,
                       volume= [randint(100,10000) for _ in range(12)]))
df

        date company  volume
0  2021-01-01    AAPL    1470
1  2021-01-01      FB    7478
2  2021-01-01    NVDA    9156
3  2021-01-01    AMZN    5972
4  2021-01-02    AAPL    9836
5  2021-01-02      FB    1990
6  2021-01-02    NVDA    5380
7  2021-01-02    AMZN    1338
8  2021-01-03    AAPL    9235
9  2021-01-03      FB    3708
10 2021-01-03    NVDA     480
11 2021-01-03    AMZN    2805

然后我创建了一个组，并且能够以与您相同的方式获取每个公司的平均值:

grp = df.groupby("company")
grp.mean()
              volume
company             
AAPL     6847.000000
AMZN     3371.666667
FB       4392.000000
NVDA     5005.333333

然后我只是将平均系列与原始 df 合并:

# on = "company" will make it align the values on the company values
merged = df.merge(grp.mean(), on= "company", suffixes= ("_daily", "_mean"))
merged
         date company  volume_daily  volume_mean
0  2021-01-01    AAPL          1470  6847.000000
1  2021-01-02    AAPL          9836  6847.000000
2  2021-01-03    AAPL          9235  6847.000000
3  2021-01-01      FB          7478  4392.000000
4  2021-01-02      FB          1990  4392.000000
5  2021-01-03      FB          3708  4392.000000
6  2021-01-01    NVDA          9156  5005.333333
7  2021-01-02    NVDA          5380  5005.333333
8  2021-01-03    NVDA           480  5005.333333
9  2021-01-01    AMZN          5972  3371.666667
10 2021-01-02    AMZN          1338  3371.666667
11 2021-01-03    AMZN          2805  3371.666667

最后，我通过做一个很好的旧减法来创建差异列:

merged["difference"] = merged["volume_daily"] - merged["volume_mean"]
merged
         date company  volume_daily  volume_mean   difference
0  2021-01-01    AAPL          1470  6847.000000 -5377.000000
1  2021-01-02    AAPL          9836  6847.000000  2989.000000
2  2021-01-03    AAPL          9235  6847.000000  2388.000000
3  2021-01-01      FB          7478  4392.000000  3086.000000
4  2021-01-02      FB          1990  4392.000000 -2402.000000
5  2021-01-03      FB          3708  4392.000000  -684.000000
6  2021-01-01    NVDA          9156  5005.333333  4150.666667
7  2021-01-02    NVDA          5380  5005.333333   374.666667
8  2021-01-03    NVDA           480  5005.333333 -4525.333333
9  2021-01-01    AMZN          5972  3371.666667  2600.333333
10 2021-01-02    AMZN          1338  3371.666667 -2033.666667
11 2021-01-03    AMZN          2805  3371.666667  -566.666667

DataFrame.merge docs
对于百分比差异:

merged["%_diff"] = merged["difference"]/merged["volume_mean"]*100
merged
         date company  volume_daily  volume_mean   difference     %_diff
0  2021-01-01    AAPL          1470  6847.000000 -5377.000000 -78.530743
1  2021-01-02    AAPL          9836  6847.000000  2989.000000  43.654155
2  2021-01-03    AAPL          9235  6847.000000  2388.000000  34.876588
3  2021-01-01      FB          7478  4392.000000  3086.000000  70.264117
4  2021-01-02      FB          1990  4392.000000 -2402.000000 -54.690346
5  2021-01-03      FB          3708  4392.000000  -684.000000 -15.573770
6  2021-01-01    NVDA          9156  5005.333333  4150.666667  82.924880
7  2021-01-02    NVDA          5380  5005.333333   374.666667   7.485349
8  2021-01-03    NVDA           480  5005.333333 -4525.333333 -90.410229
9  2021-01-01    AMZN          5972  3371.666667  2600.333333  77.123085
10 2021-01-02    AMZN          1338  3371.666667 -2033.666667 -60.316362
11 2021-01-03    AMZN          2805  3371.666667  -566.666667 -16.806723