蚂蚁面试SQL—分层贪心招聘问题

题目描述

某互联网大厂在进行员工招聘，给定固定的薪资预算【12万】，公司制定了如下贪心招聘策略：

1. 优先招聘 高职级员工（P7>P6>P5）
2. 在同一职级内，优先招聘期望薪资更低的候选人
3. 在满足高职级招聘人数尽可能多的前提下，再考虑低职级
4. 总招聘薪资不能超过预算

数据模拟

CREATE TABLE dwd_emp_candidate_di (
    candidate_id VARCHAR(20) COMMENT '候选者ID',
    level VARCHAR(20) COMMENT '应聘职级',
    salary DOUBLE COMMENT '预期薪资（单位：万）'
);

INSERT INTO dwd_emp_candidate_di VALUES
('001', 'P6', 2.5),
('002', 'P6', 2.5),
('003', 'P7', 5.0),
('004', 'P7', 4.5),
('005', 'P7', 4.0),
('006', 'P6', 3.0),
('007', 'P5', 2.0),
('008', 'P5', 1.8);

思路分析

• 本题本质上是一个 带有优先级约束的贪心选择问题
  • 首先将P7员工按照薪资排序，优先招聘工资较低的员工，确保不会超过预算，计算剩余预算
  • 然后再将P6员工按照薪资排序，优先招聘工资较低的员工，确保不会超过上述剩余预算，计算剩余预算
  • 以此类推

参考答案

WITH
p7_ranked AS (
    SELECT
        candidate_id,
        level,
        salary,
        SUM(salary) OVER (ORDER BY salary) AS running_salary,
        row_number() over (ORDER BY salary) AS rk
    FROM dwd_emp_candidate_di
    WHERE level = 'P7'
),
p7_hired AS (
    SELECT *
    FROM p7_ranked
    WHERE running_salary <= 12
),
p7_budget AS (
    SELECT COALESCE(SUM(salary), 0) AS used_budget
    FROM p7_hired
),
p6_ranked AS (
    SELECT
        candidate_id,
        level,
        salary,
        SUM(salary) OVER (ORDER BY salary) AS running_salary
    FROM dwd_emp_candidate_di
    WHERE level = 'P6'
),
p6_hired AS (
    SELECT p6.*
    FROM p6_ranked p6
    JOIN p7_budget b
    ON 1 = 1
    WHERE p6.running_salary <= (12 - b.used_budget)
),
p6_budget AS (
    SELECT COALESCE(SUM(salary), 0) AS used_budget
    FROM p6_hired
),
p5_ranked AS (
    SELECT
        candidate_id,
        level,
        salary,
        SUM(salary) OVER (ORDER BY salary) AS running_salary
    FROM dwd_emp_candidate_di
    WHERE level = 'P5'
),
p5_hired AS (
    SELECT p5.*
    FROM p5_ranked p5
    JOIN p7_budget b7
    ON 1 = 1
    JOIN p6_budget b6
    ON 1 = 1
    WHERE p5.running_salary <= (12 - b7.used_budget - b6.used_budget)
)
SELECT candidate_id, level, salary FROM p7_hired
UNION ALL
SELECT candidate_id, level, salary FROM p6_hired
UNION ALL
SELECT candidate_id, level, salary FROM p5_hired;