按日期分组找到平均不同的客户

Question

我有一个包含一个月数据的 DataFrame：

initiated_date      | specialist_id
21/10/2020 05:00:01 |   ab12
21/10/2020 12:20:01 |   gc35
22/10/2020 04:30:01 |   ad32
22/10/2020 03:40:01 |   fe45
22/10/2020 01:50:01 |   ad32
23/10/2020 02:10:01 |   iu99
23/10/2020 11:30:01 |   iu99

我想找出每天specialist_id不同的平均数名称（星期一，星期二..等）我想复制 SQL 的子查询：

SELECT 
    initiated_day, CEILING(AVG(specialist_id)) AS specialist_id
FROM
    (SELECT 
        DATE(initiated_date),
            DAYNAME(initiated_date) AS initiated_day,
            COUNT(DISTINCT specialist_id) specialist_id
    FROM
        nts.contacts
    GROUP BY 1 , 2) x
GROUP BY 1

我要找的是：

Day    |  specialist_id
Mon    |   42 
Tue    |   48
Wed    |   51
Thu    |   47
Fri    |   38
Sat    |   31
Sun    |   22

这就是我想要做的

df.groupby([df['initiated_date'].dt.date,df['initiated_date'].dt.weekday_name])['specialist_id'].nunique().reset_index()

我不确定如何更进一步。

Answer 1

您可以添加第二个groupby

st1 = dt.groupby([dt['initiated_date'].dt.date,dt['initiated_date']. day_name()])['specialist_id'].nunique()
out = st1.groupby(level=1).apply(lambda x : np.ceil(x.mean())).reset_index()

Answer 2

IIUC,

new_df = \
df.groupby(df['initiated_date'].dt.day_name())['specialist_id']\
.value_counts()\
.mean(level='initiated_date')\ #.groupby(level=0).mean() if you need instead
.rename_axis('Day').reset_index(name='specialist_id')

如果你想在白天获得独特的：

new_df = \
df.groupby(df['initiated_date'].dt.day_name())['specialist_id']\
  .nunique()\
  .rename_axis('Day').reset_index(name='specialist_id')

如果需要ceil:

new_df = \
np.ceil(
    df.groupby(df['initiated_date'].dt.day_name())['specialist_id']
      .value_counts()
      .mean(level='initiated_date')#.groupby(level=0).mean() if you need instead
)\
.rename_axis('Day').reset_index(name='specialist_id')

按日期分组找到平均不同的客户

Group by Date find average distinct customers

python

dataframe

pandas

pandas-groupby