sum(lead_num) over(partition by course_model,dy_model,origin_channel,origin_fourth_level,origin_third_level,origin_second_level,origin_first_level,lead_date order by lead_date rows between 1 preceding and 1 following)

这是一个使用窗口函数进行累计求和的 SQL 语句。该语句使用了 `SUM` 函数和 `OVER` 子句来计算每一行的 `lead_num` 值与其前后相邻的两行的 `lead_num` 值之和。

具体来说：

- `PARTITION BY` 子句按照指定的列对数据进行分组，这里是 `course_model, dy_model, origin_channel, origin_fourth_level, origin_third_level, origin_second_level, origin_first_level, lead_date`，也就是在这些列相同的行被视为同一组。
- `ORDER BY` 子句按照 `lead_date` 列对每个分组内的行进行排序，这里是按照 `lead_date` 的先后顺序进行排序。
- `ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING` 子句指定了每一行的窗口范围，即将当前行和前一行、后一行的值相加。

因此，该语句会计算每个分组内每一行的 `lead_num` 与其前后相邻两行的 `lead_num` 之和。

例如，如果我们有以下数据：

course_model | dy_model | origin_channel | origin_fourth_level | origin_third_level | origin_second_level | origin_first_level | lead_date  | lead_num
-------------|----------|------------------|-------------------------|------------------------|------------------------|----------------------|-----------|---------
course1        | dy1         | channel1           | fourth1                        | third1                       | second1                       | first1                       | 2021-01-01 | 10
course2        | dy2         | channel1           | fourth1                        | third1                       | second1                       | first1                       | 2021-01-01 | 5
course1        | dy1         | channel1           | fourth1                        | third1                       | second1                       | first1                       | 2021-01-02 | 8
course2        | dy2         | channel1           | fourth1                        | third1                       | second1                       | first1                       | 2021-01-02 | 6

则运行以下 SQL 语句：

SELECT
  course_model,
  dy_model,
  origin_channel,
  origin_fourth_level,
  origin_third_level,
  origin_second_level,
  origin_first_level,
  lead_date,
  lead_num,
  SUM(lead_num) OVER (
    PARTITION BY course_model, dy_model, origin_channel, origin_fourth_level, origin_third_level, origin_second_level, origin_first_level, lead_date
    ORDER BY lead_date
    ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING
  ) AS lead_num_sum
FROM your_table;

将得到以下结果：

course_model | dy_model | origin_channel | origin_fourth_level | origin_third_level | origin_second_level | origin_first_level | lead_date  | lead_num | lead_num_sum
-------------|----------|------------------|-------------------------|------------------------|------------------------|----------------------|-----------|-----------|-------------
course1        | dy1         | channel1           | fourth1                        | third1                       | second1                       | first1                       | 2021-01-01 | 10        | 18
course2        | dy2         | channel1           | fourth1                        | third1                       | second1                       | first1                       | 2021-01-01 | 5          | 11
course1        | dy1         | channel1           | fourth1                        | third1                       | second1                       | first1                       | 2021-01-02 | 8          | 18
course2        | dy2         | channel1           | fourth1                        | third1                       | second1                       | first1                       | 2021-01-02 | 6          | 11

其中，`lead_num_sum` 列即为每个分组内每一行的 `lead_num` 与其前后相邻两行的 `lead_num` 之和。