我在hive中有一个名为purchase_data
的表格,其中列出了所有购买的清单
我需要查询此表并查找客户购买的最昂贵产品的cust_id,product_id和价格
purchase_data
表中的数据如下所示:
cust_id product_id price purchase_data
--------------------------------------------------------
aiman_sarosh apple_iphone5s 55000 01-01-2014
aiman_sarosh apple_iphone6s 65000 01-01-2017
jeff_12 apple_iphone6s 65000 01-01-2017
jeff_12 dell_vostro 70000 01-01-2017
missy_el lenovo_thinkpad 70000 01-02-2017
我已经编写了下面的代码,但它没有获取正确的行 有些行正在重复:
select master.cust_id, master.product_id, master.price
from
(
select cust_id, product_id, price
from purchase_data
) as master
join
(
select cust_id, max(price) as price
from purchase_data
group by cust_id
) as max_amt_purchase
on max_amt_purchase.price = master.price;
输出:
aiman_sarosh apple_iphone6s 65000.0
jeff_12 apple_iphone6s 65000.0
jeff_12 dell_vostro 70000.0
jeff_12 dell_vostro 70000.0
missy_el lenovo_thinkpad 70000.0
missy_el lenovo_thinkpad 70000.0
Time taken: 21.666 seconds, Fetched: 6 row(s)
代码有问题吗?
答案 0 :(得分:0)
使用row_number()
:
select pd.*
from (select pd.*,
row_number() over (partition by cust_id order by price_desc) as seqnum
from purchase_data pd
) pd
where seqnum = 1;
每个cust_id
返回一行,即使有联系。如果您在有联系时需要多行,请使用rank()
或dense_rank()
代替row_number()
。
答案 1 :(得分:0)
我更改了代码,现在正在运行:
select master.cust_id, master.product_id, master.price
from
purchase_data as master,
(
select cust_id, max(price) as price
from purchase_data
group by cust_id
) as max_price
where master.cust_id=max_price.cust_id and master.price=max_price.price;
输出:
aiman_sarosh apple_iphone6s 65000.0
missy_el lenovo_thinkpad 70000.0
jeff_12 dell_vostro 70000.0
Time taken: 55.788 seconds, Fetched: 3 row(s)