如何基于含cumulative_sum列的账户表按月份计算汇总累计值?
需求实现方案
这个需求完全可以实现,核心思路是先补全每个账户在所有月份的记录,再匹配每个账户对应月份的最新累计值,最后按月份聚合计算总和并处理账户合并与月份标记。
实现步骤与SQL代码(以支持窗口函数的数据库为例,如MySQL 8+、PostgreSQL、SQL Server)
WITH all_month_account AS ( -- 生成所有月份与账户的全量组合,确保每个账户在每个月份都有记录 SELECT m.month, a.account FROM (SELECT DISTINCT month FROM your_table) m CROSS JOIN (SELECT DISTINCT account FROM your_table) a ), latest_cumulative AS ( SELECT ama.month, ama.account, -- 取该账户在当前月份及之前的最新累计值(因累计值递增,MAX即为最新) MAX(t.cumulative_sum) OVER ( PARTITION BY ama.account ORDER BY ama.month ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW ) AS latest_sum, -- 标记当前月份该账户是否有原始数据 CASE WHEN t.month IS NOT NULL THEN 1 ELSE 0 END AS has_original_data FROM all_month_account ama LEFT JOIN your_table t ON ama.account = t.account AND ama.month = t.month ) SELECT -- 对存在账户无原始数据的月份添加*标记 CASE WHEN SUM(1 - has_original_data) > 0 THEN CONCAT(month, '*') ELSE month END AS month, SUM(latest_sum) AS cumulative_sum, GROUP_CONCAT(DISTINCT account ORDER BY account SEPARATOR ', ') AS account FROM latest_cumulative GROUP BY month ORDER BY month;
代码逻辑说明
all_month_accountCTE:通过交叉连接所有唯一月份和唯一账户,生成每个账户对应所有月份的组合,解决部分账户在某些月份无数据的问题。latest_cumulativeCTE:左连接原表获取原始数据,用窗口函数MAX() OVER (...)提取每个账户到当前月份的最新累计值,同时标记该月份是否有原始数据。- 最终聚合查询:按月份分组,计算累计值总和,用
GROUP_CONCAT合并账户列表,对存在缺失原始数据的月份添加*标记。
兼容老版本数据库(如MySQL 5.x)的替代方案
如果数据库不支持窗口函数,可以用关联子查询实现:
SELECT -- 判断当前月份是否有账户无原始数据,有则添加* CASE WHEN EXISTS ( SELECT 1 FROM (SELECT DISTINCT month FROM your_table) m CROSS JOIN (SELECT DISTINCT account FROM your_table) a LEFT JOIN your_table t ON a.account = t.account AND m.month = t.month WHERE m.month = agg.month AND t.month IS NULL ) THEN CONCAT(agg.month, '*') ELSE agg.month END AS month, agg.cumulative_sum, agg.account FROM ( SELECT m.month, -- 子查询获取每个账户到当前月份的最新累计值 SUM( (SELECT MAX(cumulative_sum) FROM your_table t WHERE t.account = a.account AND t.month <= m.month) ) AS cumulative_sum, GROUP_CONCAT(a.account ORDER BY a.account SEPARATOR ', ') AS account FROM (SELECT DISTINCT month FROM your_table) m CROSS JOIN (SELECT DISTINCT account FROM your_table) a GROUP BY m.month ) agg ORDER BY agg.month;
内容的提问来源于stack exchange,提问作者Ganesh Khadka
相关产品推荐
相关产品推荐

