如何计算 MySQL 中收入的每日百分比变化？

Question

I have a table revenue which contains

Date_Field | Country | Category | Revenue | Buyers

I want to calculate Day on Day percentage increase in revenue. I have written the below code and am getting the correct output. But I am not sure whether the code is generic and will work for test inputs.

select a.date_field, a.curr_rev, 
case when prev_revenue is null then 'None'
else round(((curr_rev - prev_revenue)/prev_revenue*100),2) 
end as percent_change
from
(select a.date_field, a.revenue as curr_rev, b. revenue as prev_revenue from (
select date_field, sum(revenue) as revenue from revenue group by date_field) a
left join
(select date_field, sum(revenue) as revenue from revenue group by date_field) b
on a.date_field = b.date_field + interval 1 day) a
order by a.date_field

Please let me know whether the code is correct or not

Expected Output:

+------------+----------+----------------+
| date_field | curr_rev | percent_change |
+------------+----------+----------------+
| 2016-01-21 | 330      | None           |
| 2016-01-22 | 240      |-27.27          |
| 2016-01-23 | 375      | 56.25          |
+------------+----------+----------------+

Answer 1

If you are running MySQL 8.0, just use window functions:

select
    date_field,
    sum(revenue) curr_rev,
    round(
        (sum(revenue) - lag(sum(revenue)) over(order by date_field)) 
        / lag(sum(revenue)) over(order by date_field)
        * 100,
        2
    ) percent_change
from mytable

In earlier versions, you can indeed join two aggregate queries. I would phrase this as:

select
    cur.date_field,
    cur.revenue cur_rev,
    round((cur.revenue - prev.revenue) / prev.revenue * 100, 2) percent_change
from (select date_field, sum(revenue) revenue from mytable group by date_field) cur
left join (select date_field, sum(revenue) revenue from mytable group by date_field) prev 
    on prev.date_field = cur.date_field - interval 1 day

Answer 2

Assuming you have data for each day, I think you want:

select date_field,
       sum(revenue) as revenue,
       100 * ((sum(revenue) / lag(sum(revenue)) over (order by date_field)) - 1) as increase
from revenue
group by date_field;

If you don't have data on each day, then this is a little tricky. One method is the self-join you are using. Another method is a window frame:

select date_field,
       sum(revenue) as revenue,
       100 * ((sum(revenue) /
               max(sum(revenue)) over (order by date_field range between interval 1 day preceding and 1 day preceding)
              ) - 1
             ) as increase
from revenue
group by date_field;