sql - 使用窗口函数 lag() 计算百分比变化

Question

我正在尝试计算每行共享特定属性的差异。例如，我玩了三款游戏(RDR2、GTA5、Spyro)一周，我计算了每天花在每款游戏上的百分比，我想看看每天的百分比变化是多少。

因此，我写道:

percent - LAG(percent, 1) OVER(PARTITION BY game ORDER BY time DESC)AS games_delta

但不是得到期望的输出:

 |-------------------------------|
 | TIME    GAME   PERCENT  DELTA |
 |-------------------------------|
 | Dec 10  RDR2   25       15    |
 | Dec 10  GTA5   50       40    |
 | Dec 10  Spyro  25      -55    |
 | Dec 9   RDR2   10      -22    |
 | Dec 9   GTA5   10       10    |
 | Dec 9   Spyro  80       13    |
 | Dec 8   RDR2   33       ...
 | Dec 8   GTA5   0        ...
 | Dec 8   Spyro  67       ...

我得到了多个游戏并且完全错了game-delta!:

 |-------------------------------|
 | TIME    GAME   PERCENT  DELTA |
 | Dec 10  RDR2   25       2     |
 | Dec 10  RDR2   25       4     |
 | Dec 10  RDR2   25      -4     |
 | Dec 10  RDR2   25       4     |
 | Dec 10  GTA5   10       5  ...

有什么建议吗？提前致谢!

编辑:

SELECT  time,
        game,
        percent,
        games_delta

我的 GROUP BY 和 ORDER BY 中也有以下内容:

GROUP BY time, game, percent, games_delta
ORDER BY time DESC

编辑 2:

这是我的 CTE:

cte_table4 (time, games_delta) AS
(
    SELECT time,
            percent - LAG(percent, 1) OVER (PARTITION BY game ORDER BY time DESC) AS games_delta

    FROM cte_table3
)

Answer 1

在与LAG() 相同的 CTE 中添加聚合GROUP BY 将解决问题