每日HiveSQL_统计即时订单_13

1.订单配送中，如果期望配送日期和下单日期相同，称为即时订单，如果期望配送日期和下单日期不同，称为计划订单。请从配送信息表（delivery_info）中求出每个用户的首单（用户的第一个订单）中即时订单的比例，保留两位小数，以小数形式显示

需求结果

2.所用到的表和数据

--配送信息表
CREATE TABLE delivery_info
(
    `delivery_id` string comment '配送单id',
    `order_id`    string comment '订单id',
    `user_id`     string comment '用户id',
    `order_date`  string comment '下单日期',
    `custom_date` string comment '期望配送日期'
) COMMENT '邮寄信息表'
    ROW FORMAT DELIMITED FIELDS TERMINATED BY '\t';

--数据装载
insert overwrite table delivery_info
values ('1', '1', '101', '2021-09-27', '2021-09-29'),
       ('2', '2', '101', '2021-09-28', '2021-09-28'),
       ('3', '3', '101', '2021-09-29', '2021-09-30'),
       ('4', '4', '101', '2021-09-30', '2021-10-01'),
       ('5', '5', '102', '2021-10-01', '2021-10-01'),
       ('6', '6', '102', '2021-10-01', '2021-10-01'),
       ('7', '7', '102', '2021-10-01', '2021-10-03'),
       ('8', '8', '102', '2021-10-02', '2021-10-02'),
       ('9', '9', '103', '2021-10-02', '2021-10-03'),
       ('10', '10', '103', '2021-10-02', '2021-10-04'),
       ('11', '11', '103', '2021-10-02', '2021-10-02'),
       ('12', '12', '103', '2021-10-03', '2021-10-03'),
       ('13', '13', '104', '2021-10-03', '2021-10-04'),
       ('14', '14', '104', '2021-10-03', '2021-10-04'),
       ('15', '15', '104', '2021-10-03', '2021-10-03'),
       ('16', '16', '104', '2021-10-03', '2021-10-03'),
       ('17', '17', '105', '2021-10-04', '2021-10-04'),
       ('18', '18', '105', '2021-10-04', '2021-10-06'),
       ('19', '19', '105', '2021-10-04', '2021-10-06'),
       ('20', '20', '105', '2021-10-04', '2021-10-04'),
       ('21', '21', '106', '2021-10-04', '2021-10-04'),
       ('22', '22', '106', '2021-10-05', '2021-10-05'),
       ('23', '23', '106', '2021-10-05', '2021-10-05'),
       ('24', '24', '106', '2021-10-05', '2021-10-07'),
       ('25', '25', '107', '2021-10-05', '2021-10-05'),
       ('26', '26', '107', '2021-10-05', '2021-10-06'),
       ('27', '27', '107', '2021-10-06', '2021-10-06'),
       ('28', '28', '107', '2021-10-06', '2021-10-07'),
       ('29', '29', '108', '2021-10-06', '2021-10-06'),
       ('30', '30', '108', '2021-10-06', '2021-10-06'),
       ('31', '31', '108', '2021-10-07', '2021-10-09'),
       ('32', '32', '108', '2021-10-07', '2021-10-09'),
       ('33', '33', '109', '2021-10-07', '2021-10-08'),
       ('34', '34', '109', '2021-10-07', '2021-10-08'),
       ('35', '35', '109', '2021-10-08', '2021-10-10'),
       ('36', '36', '109', '2021-10-08', '2021-10-09'),
       ('37', '37', '1010', '2021-10-08', '2021-10-10'),
       ('38', '38', '1010', '2021-10-08', '2021-10-10'),
       ('39', '39', '1010', '2021-10-08', '2021-10-09'),
       ('40', '40', '1010', '2021-10-08', '2021-10-09');

3.答案

思路1:不使用rank()

3.1对同一用户同一天的下单记录且期望发货日期进行去重，保证后续rank()开窗结果排名不存在重复的排名

select user_id,
       order_date,
       custom_date
from delivery_info
group by user_id, order_date, custom_date;

运行结果

3.2rank()开窗，以user_id进行分区，order_date、custom_date进行升序排序，得出rk

select user_id,
       order_date,
       custom_date,
       rank() over (
           partition by
               user_id
           order by
               order_date,custom_date
           ) rk
from (
         select user_id,
                order_date,
                custom_date
         from delivery_info
         group by user_id, order_date, custom_date) t1;

运行结果

3.3筛选出rk = 1的记录，计算得出即时比率

select cast(
               sum(if(order_date = custom_date, 1, 0)) / count(*) as decimal(16, 2)
           ) percentage
from (
         select user_id,
                order_date,
                custom_date,
                rank() over (
                    partition by
                        user_id
                    order by
                        order_date,custom_date
                    ) rk
         from (
                  select user_id,
                         order_date,
                         custom_date
                  from delivery_info
                  group by user_id, order_date, custom_date) t1) t2
where rk = 1;

运行结果

思路2：使用row_number()

3.1第一步，使用row_number()开窗，保证每个用户的第一名只有一个

select user_id,
       order_date,
       custom_date,
       row_number() over (
           partition by
               user_id
           order by
               order_date,custom_date
           ) rn
from delivery_info;

运行结果

3.2第二步，筛选出rn = 1的记录，得出比率

select cast(
               sum(if(order_date = custom_date, 1, 0)) / count(*) as decimal(16, 2)
           ) percentage
from (
         select user_id,
                order_date,
                custom_date,
                row_number() over (
                    partition by
                        user_id
                    order by
                        order_date,custom_date
                    ) rn
         from delivery_info
     ) t1
where rn = 1;

运行结果