· 8 years ago · Nov 24, 2017, 07:20 PM
1-- DATABASE CREATION
2CREATE DATABASE IF NOT EXISTS MOVIELENS_1M_DB COMMENT "MovieLens";
3
4-- USER TABLE CREATION
5CREATE EXTERNAL TABLE IF NOT EXISTS MOVIELENS_1M_DB.USERS(
6 USER_ID INT COMMENT 'USER ID ANONYMIZED',
7 GENDER STRING COMMENT 'M=MALE, F=FEMALE',
8 AGE INT COMMENT '7 TERM AGE',
9 OCCUPATION INT COMMENT 'USER OCCUPATION CODE',
10 ZIPCODE INT COMMENT 'POSTAL CODE - US BASED'
11)
12COMMENT 'USER DEMOGRAPHIC INFO'
13-- DEFINE SERIALIZADOR ESPECÃFICO
14ROW FORMAT
15 SERDE 'org.apache.hadoop.hive.contrib.serde2.MultiDelimitSerDe'
16 WITH SERDEPROPERTIES ('field.delim'='::')
17STORED AS TEXTFILE
18LOCATION '/user/data/movielens-dataset/'
19TBLPROPERTIES ('CREATOR'='MARKU','CREATED_AT'='2017-11-16');
20
21-- MOVIES TABLE CREATION
22CREATE EXTERNAL TABLE IF NOT EXISTS MOVIELENS_1M_DB.MOVIES(
23 MOVIE_ID INT COMMENT 'MOVIE IDENTIFICATION',
24 TITLE STRING COMMENT 'MOVIE TITLE, INCLUDING YEAR',
25 GENRE ARRAY<STRING> COMMENT 'LIST OF GENRES'
26)
27COMMENT 'MOVIES LIST'
28-- DEFINE SERIALIZADOR ESPECÃFICO
29ROW FORMAT
30 SERDE 'org.apache.hadoop.hive.contrib.serde2.MultiDelimitSerDe'
31 WITH SERDEPROPERTIES ('field.delim'='::','collection.delim'='|')
32STORED AS TEXTFILE
33LOCATION '/user/data/movielens-dataset/'
34TBLPROPERTIES ('CREATOR'='MARKU','CREATED_AT'='2017-11-16');
35
36-- RATINGS TABLE CREATION
37CREATE EXTERNAL TABLE IF NOT EXISTS MOVIELENS_1M_DB.RATINGS(
38 USER_ID INT COMMENT 'USER IDENTIFICATION',
39 MOVIE_ID INT COMMENT 'MOVIE IDENTIFICATION',
40 RATING INT COMMENT 'MOVIE RATING BY USER',
41 TIMESTAMP INT COMMENT 'WHEN THE RATING WAS MADE'
42)
43COMMENT 'RATINGS TABLE - LOOKS FOR MOVIES AND USERS TABLES'
44-- DEFINE SERIALIZADOR ESPECÃFICO
45ROW FORMAT
46 SERDE 'org.apache.hadoop.hive.contrib.serde2.MultiDelimitSerDe'
47 WITH SERDEPROPERTIES ('field.delim'='::')
48STORED AS TEXTFILE
49LOCATION '/user/data/movielens-dataset/'
50TBLPROPERTIES ('CREATOR'='MARKU','CREATED_AT'='2017-11-16');