用下一个非空值填充空白

Question

从几天前开始，我一直在努力寻找解决方案。我有以下数据集。

|id|order|certain_event|order_of_occurrence|
|--|-----|-------------|-------------------|
|a |1    |NULL         |NULL               |
|a |2    |NULL         |NULL               |
|a |3    |NULL         |NULL               |
|a |4    |NULL         |NULL               |
|a |5    |4            |1                  |
|a |6    |NULL         |NULL               |
|a |7    |NULL         |NULL               |
|a |8    |4            |2                  |
|a |9    |NULL         |NULL               |

所需的输出包括用下一个非空值替换 order_of_occurrence 列中的空值。像这样：

|id|order|certain_event|order_of_occurrence|
|--|-----|-------------|-------------------|
|a |1    |NULL         |1                  |
|a |2    |NULL         |1                  |
|a |3    |NULL         |1                  |
|a |4    |NULL         |1                  |
|a |5    |4            |1                  |
|a |6    |NULL         |2                  |
|a |7    |NULL         |2                  |
|a |8    |4            |2                  |
|a |9    |NULL         |NULL               |

我尝试使用子查询从出现顺序列中检索非空值，但我得到了多个返回值。像下面这样：

SELECT a.*,
      CASE
         WHEN a.order_of_occurrence IS NOT NULL THEN a.order_of_occurence
         WHEN a.order_of_occurence IS NULL THEN (SELECT B.ORDER_OF_OCCURENCE FROM dataset AS B 
                                                 WHERE B.ORDER_OF_OCCURRENCE IS NOT NULL)
      END AS corrected_order
FROM dataset AS a

谢谢！

Answer 1

这是 FIRST/LAST_VALUE 中 IGNORE NULLS 选项的简单任务：

last_value(order_of_occurrence IGNORE NULLS)
over (partition by id
      order by "order" DESC
      rows unbounded preceding)