In hive, standard SQL is used to implement intra-group sorting in hive. To implement intra-group sorting, you generally write your own udf to implement the analysis function row_number () over (partition) in oracle, if you do not use a user-defined udf and only use standard SQL for implementation, there is no performance, just experiment. For convenience, the following statements are in oracle
In hive, standard SQL is used to implement intra-group sorting in hive. To implement intra-group sorting, you generally write your own udf to implement the analysis function row_number () over (partition) in oracle, if you do not use a user-defined udf and only use standard SQL for implementation, there is no performance, just experiment. For convenience, the following statements are in oracle
Using standard SQL in hive for intra-group sorting
In hive, to sort data in a group, you generally write a udf to implement the row_number () over (partition) function in oracle. If you do not use a udf, if you only use standard SQL, there is no performance at all, just experiment.
For convenience, the following statements are in oracle, but all of them belong to standard SQL, and can also be used in hive:
SQL code
Create table lxw_t (user_id VARCHAR2 (20 ),
Class VARCHAR2 (20 ),
Score NUMBER
);
SQL code
Insert into liuxiaowen. lxw_t VALUES ('user _ 1', 'China', '90 ');
Insert into liuxiaowen. lxw_t VALUES ('user _ 1', 'mat', '86 ');
Insert into liuxiaowen. lxw_t VALUES ('user _ 1', 'English ', '96 ');
Insert into liuxiaowen. lxw_t VALUES ('user _ 1', 'Sports ', '77 ');
Insert into liuxiaowen. lxw_t VALUES ('user _ 2', 'China', '88 ');
Insert into liuxiaowen. lxw_t VALUES ('user _ 2', 'mat', '65 ');
Insert into liuxiaowen. lxw_t VALUES ('user _ 2', 'English ', '67 ');
Insert into liuxiaowen. lxw_t VALUES ('user _ 2', 'Sports ', '98 ');
Commit;
SQL code
SELECT x. user_id, x. class, x. score, COUNT (1) AS seq
FROM (
SELECT a. *, B. score score2 FROM
Liuxiaowen. lxw_t,
Liuxiaowen. lxw_t B
WHERE a. user_id = B. user_id (+)
) X
WHERE x. score2> = x. score-sort by score from high to low
-- WHERE x. score> = x. score2 -- sort by score from low to high
Group by x. user_id, x. class, x. score
Order BY 1, 4
The results are listed from high to low as follows:
SQL code
USER_ID CLASS SCORE SEQ
User_1 English 96 1
User_1 language 90 2
User_1 mathematics 86 3
User_1 sports 77 4
User_2 sports 98 1
User_2 language 88 2
User_2 English 67 3
User_2 mathematics 65 4
The results are listed as follows:
SQL code
USER_ID CLASS SCORE SEQ
User_1 sports 77 1
User_1 mathematics 86 2
User_1 language 90 3
User_1 English 96 4
User_2 mathematics 65 1
User_2 English 67 2
User_2 language 88 3
User_2 sports 98 4