在this question中,有几种方法可以访问SELECT语句中的上一行。但是,我无法弄清楚如何有条件地做到这一点。
例如,假设我们有一个Transactions表:
customer_id purchase_date merchandise_type merchandise_name
-----------------------------------------------------------------
1 12 Apr Rice Jasmine
1 18 Apr Rice Basmati
1 19 Apr Rice Long-Grain
3 13 Apr Rice Jasmine
我想知道顾客在购买商品后多久改变了主意,预计产量是:
customer_id merchandise_name days
------------------------------------
1 Jasmine 6
1 Basmati 1
顾客1买了茉莉香米,然后6天后买了巴斯马蒂米饭,所以"天"在第一条记录中是6.以下代码能够执行此操作:
select customer_id, merchandise_name,
purchase_date - LAG(purchase_date) over (order by purchase_date) as days
from Transactions
然而,当有其他类型的商品时它不会起作用:
customer_id purchase_date merchandise_type merchandise_name
-----------------------------------------------------------------
1 12 Apr Rice Jasmine
1 13 Apr Cafe Maxwell
1 18 Apr Rice Basmati
1 19 Apr Rice Long-grain
1 19 Apr Cafe Nescafe
3 13 Apr Rice Jasmine
3 14 Apr Cafe Nescafe
是否有可能获得某个条件的前一行?类似的东西:
...
order by purchase_date
where LAG(merchandise_type) = merchandise_type
答案 0 :(得分:2)
您正在寻找的是OVER函数中的PARTITION BY子句:
select customer_id, merchandise_name,
purchase_date -
LAG(purchase_date) over (partition by customer_id, merchandise_type
order by purchase_date) as days
from Transactions
如果没有此子句,您将获得purchase_date
的任何先前值。