[英]How to change a column based on the difference between dates within a group?
This is probably a simple problem but I am quite a noob in SQL. I am using Impala.这可能是一个简单的问题,但我是 SQL 的菜鸟。我正在使用 Impala。 So I have data like this:
所以我有这样的数据:
New_ID![]() |
Date![]() |
Old_ID![]() |
---|---|---|
1 ![]() |
2020-11-14 12:41:21 ![]() |
0 ![]() |
1 ![]() |
2020-11-14 12:50:40 ![]() |
1 ![]() |
2 ![]() |
2020-10-14 15:22:00 ![]() |
1.5 ![]() |
2 ![]() |
2020-12-18 11:31:05 ![]() |
2 ![]() |
3 ![]() |
2020-11-14 12:42:25 ![]() |
3 ![]() |
Assuming that I group by New_ID, I need to check that the difference between the date and the date immediately following it (if one exists) is less that 2 months (just gonna assume that's 60 days).假设我按 New_ID 分组,我需要检查日期和紧随其后的日期(如果存在)之间的差异是否小于 2 个月(假设是 60 天)。 If the difference is greater than 2 months then I need to change the New_ID to Old_ID.
如果差异大于 2 个月,那么我需要将 New_ID 更改为 Old_ID。 If it's less than or equal to 2 months, then the New_ID can remain the same.
如果小于或等于 2 个月,则 New_ID 可以保持不变。 Essentially, I would like the new table to look like this:
本质上,我希望新表看起来像这样:
New_ID![]() |
Date![]() |
Old_ID![]() |
---|---|---|
1 ![]() |
2020-11-14 12:41:21 ![]() |
0 ![]() |
1 ![]() |
2020-11-14 12:50:40 ![]() |
1 ![]() |
1.5 ![]() |
2020-10-14 15:22:00 ![]() |
1.5 ![]() |
2 ![]() |
2020-12-18 11:31:05 ![]() |
2 ![]() |
3 ![]() |
2020-11-14 12:42:25 ![]() |
3 ![]() |
I have tried this code snippit and variations of it, but 1. I am not sure how to handle null values and 2. I keep getting a syntax error 'could not resolve column/field reference 'day' '我已经尝试过此代码片段及其变体,但是 1. 我不确定如何处理 null 值和 2. 我不断收到语法错误“无法解析列/字段引用‘天’”
SELECT New_ID, Old_ID, Date,
LAG(Date) OVER(partition by New_ID ORDER BY Date) as previous_date,
case when datediff(day, previous_date, Date)/30.0 >= 2 then Old_ID
else New_ID end as 'new_identifier'
From MYTABLE;
Any pointers/suggestions would be greatly appreciated.任何指针/建议将不胜感激。
The Impala date function is months_between()
-- and previous_date
is not recognized so you need to repeat the expression: Impala 日期 function 是
months_between()
——无法识别previous_date
,因此您需要重复表达式:
SELECT New_ID, Old_ID, Date,
LAG(Date) OVER (partition by New_ID ORDER BY Date) as previous_date,
(case when months_between(date, LAG(Date) OVER (partition by New_ID ORDER BY Date)) >= 2 then Old_ID
else New_ID
end) as new_identifier
From MYTABLE;
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.