SQL （九）分組數據（group by子句, having子句）

group by子句和having子句都是select子句

返回供應商DLL01的總產品數

select count(*) as num_prods, vend_id
from products
where vend_id = 'DLL01';

用分組把數據分爲多個邏輯組，對每個組進行聚集計算

創建分組：select語句的group by子句

select vend_id, count(*) as num_prods
from products
group by vend_id;

group by指示DBMS按照vend_id分組數據，然後對每個組再聚集計算。

注意不用group by會報錯：

select vend_id, count(*) as num_prods
from products;

過濾分組：having子句

厲害了，

• having子句非常類似於where子句，目前學過的所有類型的where子句都可以用having替代
• having也是過濾，但是過濾的是分組，而where過濾的是行
  
  

示例1

select cust_id, count(*) as orders
from orders
group by cust_id
having count(*) >=2;

如果沒有having子句，則輸出如下，所以having子句確實過濾了3個分組。這裏的過濾是基於分組聚集值，而不是特定行的值

示例2：having和where一起用

select vend_id, count(*) as num_prods
from products
where prod_price >= 4
group by vend_id
having count(*) >= 2;

看看供應商和價格情況

select vend_id, prod_price
from products;

可以看到，where語句先過濾了價格小於4的，然後再根據行數大於2來聚集

分組的排序

使用group by應該也給出order by，不要只依賴group by

沒有用order by

select order_num, count(*) as items
from orderitems
group by order_num
having count(*) >= 3;

出來的順序不知道是什麼順序

給出order by，按照items數量先排序，items相同的再按照訂單號排序

select order_num, count(*) as items
from orderitems
group by order_num
having count(*) >= 3
order by items, order_num;

select子句的順序

終於要總結他們的順序了

總結

• 雖然having和where很相似，但是要記住，having一定要和group by一起用，是用於過濾分組；而where是用於普通的行級過濾
• 掌握好where和having的差異
• 掌握好order by 和group by 的差異
• 掌握好設了select子句的順序