我正在寻找从他们的首次购买日期开始的SQL客户回购频率

Question

我正在尝试从客户的第一笔订单日期开始查找其回购价格。例如，对于2016年，在最初的1-365天中有多少客户购买了1倍的商品，有多少人购买了两次，等等。

我有一个transaction_detail表，如下所示：

txn_date Customer_ID  Transaction_Number    Sales 
1/2/2019    1           12345                $10
4/3/2018    1           65890                $20
3/22/2019   3           64453                $30
4/3/2019    4           88567                $20
5/21/2019   4           85446                $15
1/23/2018   5           89464                $40
4/3/2019    5           99674                $30
4/3/2019    6           32224                $20
1/23/2018   6           46466                $30
1/20/2018   7           56558                $30

我能够找到2016年购物的顾客以及他们在2016年回购了多少次，日期。

我需要一个查询的起点，我不确定如何在我的SQL代码中构建此逻辑。

任何帮助将不胜感激。

我正在使用以下查询：

WITH by_year
AS (SELECT
  Customer_ID,
  to_char(txn_date, 'YYYY') AS visit_year
FROM table
GROUP BY Customer_ID, to_char(txn_date, 'YYYY')),
with_first_year
AS (SELECT
  Customer_ID,
  visit_year,
  FIRST_VALUE(visit_year) OVER (PARTITION BY Customer_ID ORDER BY visit_year) AS first_year
FROM by_year),
with_year_number
AS (SELECT
  Customer_ID,
  visit_year,
  first_year,
  (visit_year - first_year) AS year_number
FROM with_first_year)
SELECT
  first_year  AS first_year,
  SUM(CASE WHEN year_number = 0 THEN 1 ELSE 0 END) AS year_0,
  SUM(CASE WHEN year_number = 1 THEN 1 ELSE 0 END) AS year_1,
  SUM(CASE WHEN year_number = 2 THEN 1 ELSE 0 END) AS year_2,
  SUM(CASE WHEN year_number = 3 THEN 1 ELSE 0 END) AS year_3,
  SUM(CASE WHEN year_number = 4 THEN 1 ELSE 0 END) AS year_4,
  SUM(CASE WHEN year_number = 5 THEN 1 ELSE 0 END) AS year_5,
  SUM(CASE WHEN year_number = 6 THEN 1 ELSE 0 END) AS year_6,
  SUM(CASE WHEN year_number = 7 THEN 1 ELSE 0 END) AS year_7,
  SUM(CASE WHEN year_number = 8 THEN 1 ELSE 0 END) AS year_8,
  SUM(CASE WHEN year_number = 9 THEN 1 ELSE 0 END) AS year_9
FROM with_year_number
GROUP BY first_year
ORDER BY first_year

Answer 1

使用窗口功能和聚合：

select cnt, count(*), min(customer_id), max(customer_id)
from (select customer_id, count(*) as cnt
      from (select td.*,
                   min(txn_date) over (partition by Customer_ID) as min_txn_date
            from transaction_detail td
           ) td
      where txn_date >= min_txn_date and txn_date < min_txn_date + interval '365' day
      group by customer_id
     ) c
group by cnt
order by cnt;

Answer 2

据我所知，您想知道在购买之日起一年或一年后于2016年首次购买并回购的不重复人士的人数。

Select * from
(
  Select customer_id, 
         Floor(months_between(txn_date, lead_txn_date)/12) as num_years
  From
  (
   Select customer_id, 
          txn_date, 
          row_number() over (partition by Customer_ID order by txn_date) as rn, 
          lead(txn_date) over (partition by Customer_ID order by txn_date) as lead_txn_date
     From your_table
  )
    Where txn_date >= date '2016-01-01' 
      and txn_date < date '2017-01-01'
      and rn = 1
      And months_between(txn_date, lead_txn_date) >= 12
)
Pivot
(
  Count(1) for num_year in (1,2,3,4)
)

最终，我们正在找到第一次和第二次购买客户之间的年数。首次购买必须在2016年。
干杯！