有点复杂：

select 
   s.*
   -- calculate rate
   , ifnull(round((s.mnt_sale - n.mnt_sale)/n.mnt_sale * 10000)/100, 0) as growth_rate
from ( 
    -- calculate monthly summary
    select month(order_date) mnt, sum(sale) mnt_sale
    from sales
    group by mnt
) s
left join ( -- join next month summary
    -- calculate monthly summary one more time
    select month(order_date) mnt, sum(sale) mnt_sale
    from sales
    group by mnt) n on n.mnt = s.mnt - 1
;

小提琴

可以使用聚合和窗口函数。类似于他的：

select year(order_date) as year, month(order_date) as month, sum(sale) as sale,
       100 * (1 - sum(sale) / lag(sum(sale), 1, sum(sale)) over (order by min(order_date)) as growth_rate
from t
group by year, month

对我来说有点棘手，但我认为下面的代码工作正常

SELECT month, sale,growth_rate
FROM(
SELECT month, sale, 
       IF(@last_entry = 0, 0, ROUND(((sale - @last_entry) / @last_entry) * 100,2)) AS growth_rate,
       @last_entry := sale AS last_entry                  
FROM
      (SELECT @last_entry := 0) x,
      (SELECT month, sum(sale) sale
       FROM   (SELECT month(order_date) as month,sum(sale) as sale 
               FROM sales GROUP BY month(order_date)) monthly_sales
       GROUP BY month) y) t;

预期结果

+-------+------+-------------+
| month | sale | growth_rate |
+-------+------+-------------+
|     1 |   90 |        0.00 |
|     2 |   65 |      -27.78 |
|     3 |  105 |       61.54 |
|     4 |   55 |      -47.62 |
+-------+------+-------------+