用于RFM分析的MySQL查询

我有一个包含RFM分析的用户/订单数据的表。 样本：

我试图写一个查询，将选择和分组的用户和金额，他们花费和分类的RFM。到目前为止，我已经：

SELECT user_id, id_o, price,
CASE WHEN price < 100 THEN '3'
     WHEN price >= 100 AND price < 500 THEN '2'
     ELSE '3'
END AS M,

CASE WHEN TIMESTAMPDIFF(DAY, o_date, DATE('2019-01-01')) < 30 THEN '1'
     WHEN TIMESTAMPDIFF(DAY, o_date, DATE('2019-01-01')) >= 30 AND
          TIMESTAMPDIFF(DAY, o_date, DATE('2019-01-01')) < 60 THEN '2'
     ELSE '3'
END AS R

FROM orders;

选择用户id、id、价格、，
如果价格<100，则为“3”
当价格>=100且价格<500时，则为“2”
其他'3'
以M结尾，
时间差（日期，o_日期，日期（'2019-01-01'））小于30，则为'1'
当时间差（日期，o_日期，日期（'2019-01-01'）>=30和
时间差（天，o_日期，日期（'2019-01-01'））<60然后是'2'
其他'3'
以R结尾
从命令；

此查询提供了一个表，其中包含每个订单、在订单上花费的金额以及标记的R和M维度。但是，我还需要添加F维度，它可以作为每个用户的订单数量来完成。我尝试将另一个CASE语句添加到查询中，并按用户id进行分组，因此如下所示：

SELECT user_id, SUM(price),
CASE WHEN price < 100 THEN '3'
     WHEN price >= 100 AND price < 500 THEN '2'
     ELSE '3'
END AS M,

CASE WHEN TIMESTAMPDIFF(DAY, o_date, DATE('2019-01-01')) < 30 THEN '1'
     WHEN TIMESTAMPDIFF(DAY, o_date, DATE('2018-01-01')) >= 30 AND
          TIMESTAMPDIFF(DAY, o_date, DATE('2018-01-01')) < 60 THEN '2'
     ELSE '3'
END AS R,

CASE WHEN COUNT(id_o) > 20 THEN '1'
     WHEN COUNT(id_o) > 10 AND COUNT(id_o) <= 20 THEN '2'
     ELSE '3'
END AS F

FROM orders GROUP BY user_id;

选择用户标识、金额（价格），
如果价格<100，则为“3”
当价格>=100且价格<500时，则为“2”
其他'3'
以M结尾，
时间差（日期，o_日期，日期（'2019-01-01'））小于30，则为'1'
时间差（日期，o_日期，日期（'2018-01-01'）>）大于等于30时，以及
时间差（天，o_日期，日期（'2018-01-01'））<60然后是'2'
其他'3'
以R结尾，
当计数（id_o）>20时，则为“1”
当COUNT（id_o）>10和COUNT（id_o）时，我将现有的工作查询转换为嵌套查询，并将附加的CASE语句放在上面，并添加了一个GROUPBY子句。这让我：

SELECT user_id, SUM(price) AS 'Total Sum', R, M, COUNT(Orders.id_o),
  CASE WHEN COUNT(Orders.id_o) < 2 THEN "3"
       WHEN COUNT(Orders.id_o) >=2 AND COUNT(id_o) <4 THEN "2"
       ELSE "1" 
  END AS F
FROM 
(SELECT user_id, id_o, price,
CASE WHEN price < 100 THEN '3'
     WHEN price >= 100 AND price < 500 THEN '2'
     ELSE '3'
END AS M,

CASE WHEN TIMESTAMPDIFF(DAY, o_date, DATE('2019-01-01')) < 30 THEN '1'
     WHEN TIMESTAMPDIFF(DAY, o_date, DATE('2019-01-01')) >= 30 AND
          TIMESTAMPDIFF(DAY, o_date, DATE('2019-01-01')) < 60 THEN '2'
     ELSE '3'
END AS R

FROM orders) AS Orders
GROUP BY user_id;

希望这对其他人也有用

SELECT user_id, SUM(price) AS 'Total Sum', R, M, COUNT(Orders.id_o),
  CASE WHEN COUNT(Orders.id_o) < 2 THEN "3"
       WHEN COUNT(Orders.id_o) >=2 AND COUNT(id_o) <4 THEN "2"
       ELSE "1" 
  END AS F
FROM 
(SELECT user_id, id_o, price,
CASE WHEN price < 100 THEN '3'
     WHEN price >= 100 AND price < 500 THEN '2'
     ELSE '3'
END AS M,

CASE WHEN TIMESTAMPDIFF(DAY, o_date, DATE('2019-01-01')) < 30 THEN '1'
     WHEN TIMESTAMPDIFF(DAY, o_date, DATE('2019-01-01')) >= 30 AND
          TIMESTAMPDIFF(DAY, o_date, DATE('2019-01-01')) < 60 THEN '2'
     ELSE '3'
END AS R

FROM orders) AS Orders
GROUP BY user_id;

+---------+-----------+------+---+--------------------+---+
| user_id | Total Sum | R    | M | COUNT(Orders.id_o) | F |
+---------+-----------+------+---+--------------------+---+
|       0 |    400.00 | 3    | 2 |                  1 | 3 |
|       1 |     90.00 | 3    | 3 |                  2 | 2 |
|      76 |     80.00 | 3    | 3 |                  2 | 2 |
+---------+-----------+------+---+--------------------+---+