当想要统计的两个指标不在一张表中时，需要做关联。但很多情况下，也没有办法保证其中一张表的维度是全的，用left join或right join可能会导致数据丢失。所以借助full join处理。

1）如，将下面的数据处理成表格中的效果（维度就是name）。

目标效果：

select if(r.name is null ,r.name2,r.name) as name, -- 保留有值的name
       r.number,
       r.show_number
from 
(   -- 数据合并
    select t1.name,t1.number,t2.name as name2,t2.show_number
     from 
        (select 'A' as name,300 as number
            union all select 'B' ,100
            union all select 'C' ,150 
        ) as t1
        full join 
        (select 'A' as name,0.8 as show_number
            union all select 'D',0.66
            union all select 'C',0.2
        ) as t2 on t1.name=t2.name
) as r

结果 ：

2）如果再增加一个维度type的话：

select if(r.name is null ,r.name2,r.name) as name, -- 保留有值的name
       if(r.type is null ,r.type2,r.type) as type,
       r.number,
       r.show_number
    --   r.type2
from 
(   -- 数据合并
    select t1.name,t1.number,t1.type,t2.name as name2,t2.show_number,t2.type as type2
     from 
        (select 'A' as name,300 as number,'red' as type
            union all select 'B' ,100,'red'
            union all select 'C' ,150 ,'red'
        ) as t1
        full join 
        (select 'A' as name,0.8 as show_number,'red' as type
            union all select 'D',0.66,'green'
            union all select 'C',0.2,'green'
        ) as t2 on t1.name=t2.name and t1.type=t2.type
) as r

结果：