SQL Sum count of product by Member and product YTD of the product count [关闭]

Question

SELECT DISTINCT 
  DMC.MEM_NBR, 
  FIRST_NAME, 
  LAST_NAME, 
  PRODUCT_DESC, 
  DD.YEAR_MONTH AS FILL_DATE, 
  FC.PRODUCT_NBR 
  CASE 
    WHEN CAST ((CURRENT_DATE - DOB.DATE_DATE) / 365.25 as Int) >= 18  Then '18+'
    ELSE 'Less than 18'
  END MEM_AGE
FROM FT_PRODUCT_CLAIM FC
LEFT JOIN DIM_PRODUCT PRODUCT ON FC.PRODUCT_DIM_CK=PRODUCT.PRODUCT_DIM_CK
LEFT JOIN DIM_MEMBER DMC ON DMC.MEMBER_DIM_CK = FC.MEMBER_DIM_CK 
INNER JOIN DIM_DATE DD ON DD.DATE_DIM_CK = FC.PRODUCT_DATE_DIM_CK 
INNER JOIN DIM_DATE DOB ON DOB.DATE_DIM_CK = DMC.BIRTH_DATE_DIM_CK 
WHERE FC.PRODUCT_NBR IN (SELECT * FROM TBL_PRODUCT)
AND FILL_DATE between 202001 and 202002
AND MEM_AGE = '18+'
ORDER BY FIRST_NAME

-- 问题： ---通过产品编号计算有多少产品与会员相关联 ---想总结每个成员的唯一产品NBRS的数量（MBR_NBR） ---然后为每个新的月份（Fill_Date）为会员的每个产品总和添加一个 YTD

Answer 1

我确信有一种更清洁的方法可以做到这一点，但这可能会奏效：

SELECT 
  src.mem_nbr, 
  MAX(src.first_name) AS first_name,
  MAX(src.last_name) AS last_name,
  MAX(src.product_desc) AS product_desc,
  src.year_month AS fill_date, 
  MAX(src.cum_cnt_yearly) AS ytd_per_month,
  MAX(uc.uniq_prd_cnt) AS uniq_prd_cnt
FROM (
  SELECT dmc.mem_nbr, dd.year_month, dmc.first_name, dmc.last_name, p.product_desc,
    CASE 
      WHEN CAST ((CURRENT_DATE - dob.date_date) / 365.25 AS INTEGER) >= 18  THEN '18+' 
      ELSE 'Less than 18' 
    END AS mem_age,
    COUNT(fc.product_nbr) OVER(
      PARTITION BY dmc.mem_nbr, <dd.year>
      ORDER BY dd.year_month
      ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW
    ) cum_cnt_yearly -- cumulative yearly row count
  FROM ft_product_claim fc -- product claims
  LEFT JOIN dim_product p ON fc.product_dim_ck = p.product_dim_ck -- product
  LEFT JOIN dim_member dmc ON dmc.member_dim_ck = fc.member_dim_ck -- member
  INNER JOIN dim_date dd ON dd.date_dim_ck = fc.product_date_dim_ck -- prod date
  INNER JOIN dim_date dob ON dob.date_dim_ck = dmc.birth_date_dim_ck -- birth date
  INNER JOIN tbl_product pd ON fc.product_nbr = pd.<prod_nbr> -- valid product
  WHERE year_month BETWEEN 202001 AND 202002
  AND mem_age = '18+'
) src
LEFT JOIN (
  SELECT dmc.mem_nbr, COUNT(DISTINCT fc.product_nbr) uniq_prd_cnt
  FROM ft_product_claim fc -- product claims
  INNER JOIN dim_member dmc ON fc.member_dim_ck = dmc.member_dim_ck -- member info
  GROUP BY dmc.mem_nbr
) uc ON src.mem_nbr = uc.mem_nbr -- unique count
GROUP BY src.mem_nbr, src.fill_date;

假设：

• 您想获取ft_product_claim 中的行数
• 您的所有联接都是 1-1 联接
• tbl_product 有一个匹配的 <prod_nbr> 字段
• dim_date 有一个 <dd.year> 字段，否则您必须使用表达式来提取它

几个观察：

1. 除非你需要所有字段，否则你不应该使用SELECT *，在这种情况下你不需要；我什至不知道它是否会按原样工作。我将WHERE FC.PRODUCT_NBR IN (...) 转换为INNER JOIN。

2. 如果您知道dim_product 和dim_member 中始终有匹配的行，请将这些连接更改为INNER JOINs。