Hive学习—行列转换

智者的梦再美，也不如愚人实干的脚印。供学习HSQL的童鞋们参考！

在工作学习中，往往需要对查询的表结构进行简单的行转列或列转行的优化

列转行

表数据如图,表名customer_details

要求：查询每个国家，女性，男性的人数，如图

我们很容易通过以下语句查出结果，但是格式和目标还需要转换下

select country,gender,count(*) as people from customer_details group by country,gender;

方法一（易于理解）

查出每个国家的男性人数作为临时表1，查出每个国家的女性人数作为临时表2，然后通过country字段来inner join实现左右连接

临时表1

select country,count(*) as female from customer_details where gender='Female' group by country

临时表2

select country,count(*) as female from customer_details where gender='Female' group by country

连接

select a.country,female,male from
(select country,count(*) as female from customer_details where gender='Female' group by country)a 
join 
(select country,count(*) as male from customer_details where gender='Male' group by country)b 
on
a.country=b.country

方法二（效率高）

利用sum,if函数计算分组时男性，女性的人数

select country,
sum(if(gender='Male',1,0)) as male,
sum(if(gender='Female',1,0)) as female
from customer_details
group by country

方法三（冷门）

分组，利用collect_list集合的size方法获取性别的数量

select a.country,female,male from
(select country,size(collect_list(gender)) as female from customer_details where gender='Female' group by country)a 
join 
(select country,size(collect_list(gender)) as male from customer_details where gender='Male' group by country)b 
on
a.country=b.country

为方便演示行转列，将上述的查询结果作为临时视图 tmpview

行转列

主要是利用union/union all实现上下连接

select country,'Female' as gender, female as people from tmpview 
union all
select country,'Male' as gender,male as people from tmpview