MySQL窗口函数实战：从基础排名到高级分析

1. MySQL窗口函数核心概念解析

窗口函数（Window Function）是MySQL 8.0引入的重要特性，它能在不减少原表行数的情况下，对数据进行分组计算和排序操作。与GROUP BY聚合不同，窗口函数会保留原始数据行的完整性，同时附加计算结果。

关键区别：传统聚合函数会将多行合并为一行，而窗口函数为每行返回一个值，同时保持原表结构不变。

窗口函数语法结构包含三个核心部分：

sql复制函数名(参数) OVER (
    [PARTITION BY 分组字段]
    [ORDER BY 排序字段]
    [frame_clause]
)

2. 排名类窗口函数实战

2.1 基础排名函数对比

MySQL提供三种主要排名函数，它们在处理相同值时的行为不同：

实际应用案例：学生成绩排名

sql复制SELECT 
    student_name,
    score,
    ROW_NUMBER() OVER(ORDER BY score DESC) AS row_num,
    RANK() OVER(ORDER BY score DESC) AS rank_val,
    DENSE_RANK() OVER(ORDER BY score DESC) AS dense_rank_val
FROM exam_results;

2.2 高级排名技巧

2.2.1 分组排名

在部门内计算员工薪资排名：

sql复制SELECT 
    department_id,
    employee_name,
    salary,
    RANK() OVER(PARTITION BY department_id ORDER BY salary DESC) AS dept_rank
FROM employees;

2.2.2 百分比排名

使用PERCENT_RANK()计算相对位置：

sql复制SELECT 
    product_id,
    sales,
    PERCENT_RANK() OVER(ORDER BY sales) AS percentile
FROM products;

注意事项：PERCENT_RANK()返回值为0到1之间，表示当前行的相对位置

3. 聚合类窗口函数深度应用

3.1 滑动窗口计算

通过frame_clause定义窗口范围，实现移动平均等计算：

sql复制SELECT 
    date,
    revenue,
    AVG(revenue) OVER(
        ORDER BY date 
        ROWS BETWEEN 2 PRECEDING AND CURRENT ROW
    ) AS moving_avg
FROM daily_sales;

常用帧规范选项：

• ROWS BETWEEN N PRECEDING AND M FOLLOWING
• RANGE BETWEEN INTERVAL 7 DAY PRECEDING AND CURRENT ROW

3.2 累计聚合分析

计算累计求和与累计百分比：

sql复制SELECT 
    month,
    sales,
    SUM(sales) OVER(ORDER BY month) AS running_total,
    sales/SUM(sales) OVER() * 100 AS percent_of_total
FROM monthly_sales;

4. 性能优化与实战技巧

4.1 索引设计策略

窗口函数性能关键点：

1. PARTITION BY字段应建立索引
2. ORDER BY字段应建立索引
3. 复合索引顺序：(分区字段, 排序字段)

4.2 执行计划解读

使用EXPLAIN分析窗口函数查询：

• 关注"Using filesort"警告
• 窗口函数通常会在最后一步执行
• 大数据集考虑添加LIMIT条件

4.3 常见问题解决方案

4.3.1 内存溢出处理

当出现"Memory exhausted"错误时：

1. 增加sort_buffer_size参数
2. 减少PARTITION BY的分组数量
3. 分批次处理数据

4.3.2 结果验证技巧

验证窗口函数结果的可靠方法：

sql复制-- 对比传统JOIN方式与窗口函数结果
WITH window_result AS (
    SELECT id, RANK() OVER(ORDER BY score) AS rnk
    FROM students
),
join_result AS (
    SELECT s.id, COUNT(DISTINCT s2.score) AS rnk
    FROM students s
    JOIN students s2 ON s.score <= s2.score
    GROUP BY s.id
)
SELECT * FROM window_result w
LEFT JOIN join_result j ON w.id = j.id
WHERE w.rnk != j.rnk;

5. 复杂业务场景综合案例

5.1 电商用户行为分析

计算用户访问的会话划分与页面停留排名：

sql复制SELECT 
    user_id,
    page_url,
    visit_time,
    TIMESTAMPDIFF(SECOND, LAG(visit_time) OVER(PARTITION BY user_id ORDER BY visit_time), visit_time) AS time_diff,
    CASE WHEN TIMESTAMPDIFF(MINUTE, LAG(visit_time) OVER(PARTITION BY user_id ORDER BY visit_time), visit_time) > 30 
         OR LAG(visit_time) OVER(PARTITION BY user_id ORDER BY visit_time) IS NULL
         THEN 1 ELSE 0 END AS new_session,
    SUM(CASE WHEN TIMESTAMPDIFF(MINUTE, LAG(visit_time) OVER(PARTITION BY user_id ORDER BY visit_time), visit_time) > 30 
              OR LAG(visit_time) OVER(PARTITION BY user_id ORDER BY visit_time) IS NULL
              THEN 1 ELSE 0 END) 
        OVER(PARTITION BY user_id ORDER BY visit_time) AS session_id,
    ROW_NUMBER() OVER(PARTITION BY user_id, session_id ORDER BY visit_time) AS page_sequence
FROM user_page_visits;

5.2 金融交易风控模型

识别异常交易模式：

sql复制SELECT 
    transaction_id,
    account_id,
    amount,
    transaction_time,
    AVG(amount) OVER(
        PARTITION BY account_id
        ORDER BY transaction_time
        RANGE BETWEEN INTERVAL 7 DAY PRECEDING AND CURRENT ROW
    ) AS avg_7day,
    amount - AVG(amount) OVER(
        PARTITION BY account_id
        ORDER BY transaction_time
        RANGE BETWEEN INTERVAL 7 DAY PRECEDING AND CURRENT ROW
    ) AS deviation,
    CASE WHEN amount > 3 * STD(amount) OVER(
                        PARTITION BY account_id
                        ORDER BY transaction_time
                        RANGE BETWEEN INTERVAL 30 DAY PRECEDING AND CURRENT ROW
                    )
         THEN 1 ELSE 0 END AS is_outlier
FROM transactions;

6. 进阶技巧与最佳实践

6.1 多窗口函数组合

高效计算多个指标：

sql复制SELECT 
    product_id,
    month,
    sales,
    RANK() OVER(PARTITION BY month ORDER BY sales DESC) AS monthly_rank,
    RANK() OVER(PARTITION BY product_id ORDER BY month) AS growth_rank,
    SUM(sales) OVER(PARTITION BY product_id ORDER BY month) AS ytd_sales,
    sales - LAG(sales) OVER(PARTITION BY product_id ORDER BY month) AS mom_growth
FROM product_sales;

6.2 动态窗口大小

根据业务规则调整窗口范围：

sql复制SELECT 
    date,
    metric,
    AVG(metric) OVER(
        ORDER BY date
        ROWS BETWEEN 
            CASE WHEN DAYOFWEEK(date) = 1 THEN 6 ELSE 1 END PRECEDING 
            AND CURRENT ROW
    ) AS dynamic_avg
FROM metrics;

6.3 窗口函数与CTE结合

复杂分析查询的可读性优化：

sql复制WITH monthly_stats AS (
    SELECT 
        department_id,
        month,
        SUM(sales) AS total_sales,
        COUNT(*) AS transaction_count
    FROM sales_data
    GROUP BY department_id, month
),
window_calc AS (
    SELECT 
        department_id,
        month,
        total_sales,
        AVG(total_sales) OVER(PARTITION BY department_id ORDER BY month ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) AS moving_avg,
        RANK() OVER(PARTITION BY month ORDER BY total_sales DESC) AS dept_rank
    FROM monthly_stats
)
SELECT * FROM window_calc WHERE dept_rank <= 3;

专业建议：对于复杂窗口函数查询，使用CTE分步处理可以显著提高可维护性

窗口函数是SQL分析的强大工具，但需要特别注意：

1. MySQL 8.0以下版本不支持
2. 大数据量查询可能消耗大量内存
3. 正确设计PARTITION BY和ORDER BY子句对性能至关重要
4. 结果验证是保证计算准确性的关键步骤