五個統計函數(單獨使用,意義不大,經常和分組group by組合使用)
max 最大select max(shop_price) from goods;
min 最小select min(shop_price) from goods;
sum 求和select sum( shop_price) from goods;
avg 求平均值select avg(shop_price) from goods;
count 所有值得行數有多少行
count(*)絕對行數null也計算在內
除此之外count(列名) ,計算這一列非null的行數
count使用
mysql> select * from test8; +------+------+ | id | name | +------+------+ | 1 | lisi | | 2 | NULL | +------+------+ mysql> select count(*) from test8; +----------+ | count(*) | +----------+ | 2 | +----------+ mysql> select count(name) from test8; +-------------+ | count(name) | +-------------+ | 1 | +-------------+
查詢類型為4的庫存
select sum(goods_number) from goods where cat_id=4;
group by
統計一下每個類型分組下的庫存
mysql> select cat_id,sum(goods_number) from goods group by cat_id; +--------+-------------------+ | cat_id | sum(goods_number) | +--------+-------------------+ | 2 | 0 | | 3 | 203 | | 4 | 4 | | 5 | 8 | | 8 | 61 | | 11 | 23 | | 13 | 4 | | 14 | 9 | | 15 | 2 | +--------+-------------------+
統計一下每個類型分組下的庫存
語上解釋不通(每個類別cat_id裡面有很多goods_name)
不推薦select goods_name ,sum(goods_number) from goods group by cat_id;
解釋:在select a/b中必須在group by a/b/c語意上才沒有問題
技巧:查詢語句理解上從後面的條件過濾開始,先理解過濾條件,再看前面的執行