Question

我正在尝试使用Postgresql查询创建RFM分析。但是，我还没有完全完成对Recency维度的查询。该文章的灵感来自于本文 “https://cooldata.wordpress.com/2014/03/25/an-all-sql-way-to-automate-rfm-scoring/”
新近度维度的标准是

2个月内的最后订单= 5
4个月内的最后订单= 4
6个月内的最后订单= 3
8个月内的最后订单= 2
10个月内的最后订单= 1

以下是我一直在努力完成的查询

WITH rfm AS

(SELECT email,
 SUM((total_incl_tax)) AS cash,
 MAX(decode(order_order.order_date, 2016-01-01, 5, 2016-02-01, 4, 2016-03-01, 3, 2016-04-01, 2, 201605-01, 1)) AS recency,
 COUNT(DISTINCT(order_date)) AS frequency
 FROM order_order

 GROUP BY email)


SELECT rfm.email,
CASE
WHEN rfm.cash >= 2000000 THEN 5
WHEN rfm.cash > 1500000 THEN 4
WHEN rfm.cash > 1000000 THEN 3
WHEN rfm.cash > 500000 THEN 2
WHEN rfm.frequency > 4 THEN 5
WHEN rfm.frequency = 4 THEN 4
WHEN rfm.frequency = 3 THEN 3
WHEN rfm.frequency = 2 THEN 2
WHEN rfm.frequency = 1 THEN 1
else 1

END  + rfm.frequency AS rfm_score
--+ Five_years.recency

FROM rfm
GROUP BY rfm.email, rfm.cash,rfm.frequency
ORDER BY rfm.email

错误是：

ERROR: function decode(timestamp with time zone, integer, integer, integer, integer, integer, integer, integer, integer, integer, integer) does not exist Hint: No function matches the given name and argument types. You might need to add explicit type casts. Position: 186

我假设错误在这一行

MAX(decode(order_order.order_date, 2016-01-01, 5, 2016-02-01, 4, 2016-03-01, 3, 2016-04-01, 2, 2016-05-01, 1)) AS recency

是否有任何建议将错误行修改为新近维度所述的标准？感谢

Answer 1

Postgres中没有decode()功能。您可以使用另一个CASE语句替换它：

WITH rfm AS
(
     SELECT email,
     SUM((total_incl_tax)) AS cash,
     MAX(
         CASE
          WHEN order_order.order_date = '2016-01-01' THEN 5
          WHEN order_order.order_date = '2016-02-01' THEN 4
          WHEN order_order.order_date = '2016-03-01' THEN 3
          WHEN order_order.order_date = '2016-04-01' THEN 2
          WHEN order_order.order_date = '2016-05-01' THEN 1
         END
        )   as recency,
     COUNT(DISTINCT(order_date)) AS frequency
     FROM order_order
     GROUP BY email
 )
SELECT rfm.email,
CASE
    WHEN rfm.cash >= 2000000 THEN 5
    WHEN rfm.cash > 1500000 THEN 4
    WHEN rfm.cash > 1000000 THEN 3
    WHEN rfm.cash > 500000 THEN 2
    WHEN rfm.frequency > 4 THEN 5
    WHEN rfm.frequency = 4 THEN 4
    WHEN rfm.frequency = 3 THEN 3
    WHEN rfm.frequency = 2 THEN 2
    WHEN rfm.frequency = 1 THEN 1
    else 1
END  + rfm.frequency + rfm.recency AS rfm_score
FROM rfm
GROUP BY rfm.email, rfm.cash,rfm.frequency
ORDER BY rfm.email

进一步阅读：Decode equivalent in postgres

使用Postgresql进行RFM分析

1 个答案: