HIVE卡口流量需求分析

1.创建表格插入数据

2.需求

3.总结：

1.创建表格插入数据

CREATE TABLE learn3.veh_pass(
id STRING COMMENT "卡口编号",
pass_time STRING COMMENT "进过时间",
pass_num int COMMENT "过车数"
)
ROW FORMAT DELIMITED FIELDS TERMINATED BY ","
STORED AS TEXTFILE;

load data local inpath "/usr/local/soft/hive-3.1.2/data/veh_pass.txt" INTO TABLE learn3.veh_pass;

2.需求

需求1：查询四月的设备及其设备种类总数

(如果是查询当前月可以使用语句：substr(pass_time,1,7) = substr(current_date,1,7))


-- 写法1
SELECT
T.id
,count(*) OVER()
FROM (
SELECT
id 
,pass_time
FROM learn3.veh_pass
WHERE substr(pass_time,1,7) = substr(current_date,1,7)
) T GROUP BY T.id


-- 错误写法
SELECT
DISTINCT id 
,count(*) OVER()
FROM (
SELECT
id 
,pass_time
FROM learn3.veh_pass
WHERE substr(pass_time,1,7) = substr(current_date,1,7)
)T 


-- 写法2：
SELECT
T1.id
,count(*) OVER()
FROM (
SELECT
DISTINCT T.id
FROM (
SELECT
id 
,pass_time
FROM learn3.veh_pass
WHERE substr(pass_time,1,7) = "2022-04"
)T )T1;

如果求的是四月每辆车的出现次数

select
t1.id
,count(*)
from
(
select
v.id
,v.pass_time
from learn3.veh_pass v
where substr(pass_time,1,7) = "2022-04"
) t1 group by t1.id;



+---------------------+-----------------+
|        t1.id        | count_window_0  |
+---------------------+-----------------+
| 451000000000071117  | 5               |
| 451000000000071116  | 5               |
| 451000000000071115  | 5               |
| 451000000000071114  | 5               |
| 451000000000071113  | 5               |
+---------------------+-----------------+

+---------------------+
|         id          |
+---------------------+
| 451000000000071113  |
| 451000000000071114  |
| 451000000000071115  |
| 451000000000071116  |
| 451000000000071117  |
+---------------------+

3.总结：

OVER():会为每条数据都开启一个窗口，默认窗口大小就是当前数据集的大小

OVER(PARTITION BY)会按照指定的字段进行分区，在获取一条数据时，窗口大小为整个分区的大小，之后根据分区中的数据进行计算

OVER(PARTITION BY ... ORDER BY ...)根据给定的分区，在获取一条数据时，窗口大小为整个分区的大小，并且对分区中的数据进行排序

-- 需求2：查询所有流量明细及所有设备月流量总额

SELECT
T1.id
,T1.pass_time
,T1.pass_num
,SUM(T1.pass_num) OVER(PARTITION BY SUBSTRING(T1.pass_time,1,7)) as total_pass
FROM learn3.veh_pass T1;

需求3：按设备编号日期顺序展示明细并求

OVER()：指定分析函数工作的数据窗口大小，这个数据窗口大小可能会随着行的改变而变化。
CURRENT ROW：当前行
n PRECEDING：往前n行数据
n FOLLOWING：往后n行数据
UNBOUNDED ：起点，
UNBOUNDED PRECEDING 表示从前面的起点，
UNBOUNDED FOLLOWING 表示到后面的终点

假设我们现在要取当前行当前行的前一行数据和后一行数据我们可以写、

ROW BETWEEN 1 PRECEDING and 1 FOLLOWING

1）从第一天开始到当前天数对流量进行累加

SELECT
T1.*
,SUM(T1.pass_num) OVER(ORDER BY T1.pass_time ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
FROM (
SELECT 
*
FROM learn3.veh_pass ORDER BY pass_time
) T1;

2）昨天与当前天流量累加


SELECT
T1.*
,SUM(T1.pass_num) OVER(ORDER BY T1.pass_time ROWS BETWEEN 1 PRECEDING AND CURRENT ROW)
FROM (
SELECT 
*
FROM learn3.veh_pass ORDER BY pass_time
) T1;

3）当前天数的前一天与后一天流量累加

SELECT
T1.*
,SUM(T1.pass_num) OVER(ORDER BY T1.pass_time ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING)
FROM (
SELECT 
*
FROM learn3.veh_pass ORDER BY pass_time
) T1;

4）当前天与下一天的累加和

SELECT
T1.*
,SUM(T1.pass_num) OVER(ORDER BY T1.pass_time ROWS BETWEEN CURRENT ROW  AND 1 FOLLOWING)
FROM (
SELECT 
*
FROM learn3.veh_pass ORDER BY pass_time
) T1;

5）当前天数与之后所有天流量累加和

SELECT
T1.*
,SUM(T1.pass_num) OVER(ORDER BY T1.pass_time ROWS BETWEEN CURRENT ROW  AND UNBOUNDED FOLLOWING)
FROM (
SELECT 
*
FROM learn3.veh_pass ORDER BY pass_time
) T1;

需求4：查询每个设备编号上次有数据日期和下一次有数据日期

LAG(col,n,default_val)：往前第n行数据
LEAD(col,n, default_val)：往后第n行数据
NTILE(n)：把有序窗口的行分发到指定数据的组中，各个组有编号，编号从1开始，对于每一行，NTILE返回此行所属的组的编号。

SELECT
T1.*
, LAG(T1.pass_time,1,"2022-01-01") OVER(PARTITION BY T1.id ORDER BY T1.pass_time) as before_time
, LEAD(T1.pass_time,1,"2022-12-31") OVER(PARTITION BY T1.id ORDER BY T1.pass_time) as after_time
FROM learn3.veh_pass T1;