差集定义:一般地,设A,B是两个集合,由所有属于A且不属于B的元素组成的集合,叫做集合A减集合B(或集合A与集合B之差)。

                   类似地,对于集合A,B,我们把集合{x/x∈A,且x¢B}叫做A与B的差集,记作A-B记作A-B(或A\B);

                   即A-B={x|x∈A,且x ¢B}(或A\B={x|x∈A,且x ¢B} B-A={x/x∈B且x¢A} 叫做B与A的差集。


比如说有这么两个表:

hive> select * from A;
OK
1 2
1 3
2 1
2 3
3 1
Time taken: 0.3 seconds, Fetched: 5 row(s)
hive> select * from B;
OK
1 2
1 4
2 2
2 3
Time taken: 0.086 seconds, Fetched: 4 row(s)



要取出A与B的差集(A-B):

1 3
2 1
3 1



Hive可不可以用not in?可以,但只能用于单个字段。select * from A where (uid,goods) not in (select uid,goods from B);这个oracle是支持的,但hive不行。

hive> select * from A  where uid not in (select uid from B);
3 1
Time taken: 46.09 seconds, Fetched: 1 row(s)



Hive可不可以用not exists?显然也可以! 

hive> select * from A  where not exists (select * from B where A.uid=B.uid and A.goods=B.goods);
1 3
2 1
3 1
Time taken: 12.989 seconds, Fetched: 3 row(s)



不过前两种貌似很费资源,在ODPS里都有限制,下面来介绍一下hive常用的求差集方法,左(右)连接 left outer join


先看一下左连接之后表是什么样的

hive> select * from A a left outer join B b on a.uid=b.uid and a.goods=b.goods;
1 2 1 2
1 3 NULL NULL
2 1 NULL NULL
2 3 2 3
3 1 NULL NULL
Time taken: 12.735 seconds, Fetched: 5 row(s)



现在只要取出B的uid和goods为null的行就可以了


hive> select a.* from A a left outer join B b on a.uid=b.uid and a.goods=b.goods where b.uid is null and b.goods is null;
1 3
2 1
3 1
Time taken: 13.023 seconds, Fetched: 3 row(s)