Question

select week(cd.date_time,4) as week, year(cd.date_time) as year, p.product_name as product_name, p.asin as asin, 
round(avg(cd.bsr_lv1)) as bsr_lv1, round(avg(cd.bsr_lv2)) as bsr_lv2, cd.stock_status as stock_status, COALESCE(NULLIF(round(avg(cd.price)),''),NULLIF(round(avg(cd.sale_price)),'' ),NULLIF(round(avg(cd.deal_price)),''), 'OOS') as price, round(avg(cd.review_total)) as reviews, round(avg(cd.rating),1) as rating,
(case when (round(sum(sr.unit_sold)) in ('','N/A',NULL)) THEN 0 ELSE round(sum(sr.unit_sold)) END) as unit_sold 
from crawl_data cd 
left join products p on p.id=cd.product_id 
left join sale_report sr on sr.product_id=cd.product_id 
where (date(cd.date_time) BETWEEN CURRENT_DATE - INTERVAL '6' WEEK AND CURRENT_DATE) and 
p.asin='B07H66KQ1B' and week(cd.date_time,4)=week(sr.date_time,4) 
group by week(cd.date_time,4), year(cd.date_time)

上面的查询产生了unit_sold的错误总和。

这里有人可以帮助我查询吗？我会非常感激。

Answer 1

问题出在案例陈述中。

(case when (round(sum(sr.unit_sold)) in ('','N/A',NULL)) THEN 0 ELSE round(sum(sr.unit_sold)) END) as unit_sold.

整数和和应该是最外面的函数。案例说明应检查单个列，而不是汇总和四舍五入。

Answer 2

这是您的查询：

select week(cd.date_time, 4) as week,
       year(cd.date_time) as year,
       p.product_name as product_name,
       p.asin as asin, 
       round(avg(cd.bsr_lv1)) as bsr_lv1,
       round(avg(cd.bsr_lv2)) as bsr_lv2,
       cd.stock_status as stock_status, 
       coalesce(nullif(round(avg(cd.price)), ''),  
                nullif(round(avg(cd.sale_price)), '' ),  
                nullif(round(avg(cd.deal_price)), ''),
                'OOS'
               ) as price,
       round(avg(cd.review_total)) as reviews, 
       round(avg(cd.rating), 1) as rating,
       (case when (round(sum(sr.unit_sold)) in ('', 'N/A', NULL)) then 0 else round(sum(sr.unit_sold)) end) as unit_sold 
from crawl_data cd left join
     products p
     on p.id = cd.product_id left join
     sale_report sr
     on sr.product_id = cd.product_id 
where date(cd.date_time) BETWEEN CURRENT_DATE - INTERVAL '6' WEEK AND CURRENT_DATE and 
      p.asin = 'B07H66KQ1B' and 
      week(cd.date_time, 4) = week(sr.date_time, 4) 
group by week(cd.date_time, 4), year(cd.date_time);

它有几个问题：

不需要产品的left join。 WHERE子句将其转换为内部联接。
select中有多个列，group by中没有。。。例如product_name，asin和stock_status。
price的计算是没有意义的，因为它是将数字和字符串组合在一起的。
units_sold的计算没有意义，因为它正在将数字函数的输出与字符串进行比较。
IN (. . . NULL)是没有用的。它将永远不会返回true。

鉴于所有这些问题，没有任何真正的解释，我只能推测一个有用的查询是什么。以下是我的最佳猜测：

select week(cd.date_time, 4) as week,
       year(cd.date_time) as year,
       p.product_name as product_name,
       p.asin as asin, 
       round(avg(cd.bsr_lv1)) as bsr_lv1,
       round(avg(cd.bsr_lv2)) as bsr_lv2,
       cd.stock_status as stock_status, 
       coalesce(nullif(round(avg(cd.price)), 0),  
                nullif(round(avg(cd.sale_price)), 0),  
                nullif(round(avg(cd.deal_price)),
               ) as price,  -- just return `NULL` if there is no price
       round(avg(cd.review_total)) as reviews, 
       round(avg(cd.rating), 1) as rating,
       round(sum(case when sr.unit_sold in ('', 'N/A') then 0 else 0 + sr.unit_sold end), 1) as unit_sold  -- is units_sold really a string???
from crawl_data cd join
     products p
     on p.id = cd.product_id left join  -- this could probably be an inner join as well
     sale_report sr
     on sr.product_id = cd.product_id 
where date(cd.date_time) BETWEEN CURRENT_DATE - INTERVAL '6' WEEK AND CURRENT_DATE and 
      p.asin = 'B07H66KQ1B' and 
      week(cd.date_time, 4) = week(sr.date_time, 4) 
group by week(cd.date_time, 4), year(cd.date_time),
         p.product_name, p.asin, cd.stock_status;

MySQL查询使用联接获取每周销售的产品总和，以从另一个表中获取平均数据

2 个答案: