sql - SQL 中的年度队列分析

Question

我正在尝试进行同类群组分析 - 尝试计算留存率。这是我的代码:

SELECT 
  renter_id,
  min(DATE_PART('year', created_at)) AS first_rental_year,
  DATE_PART('year', created_ay) AS y1,
  round(100 * count(distinct b2.renter_id) /
    count(distinct b1.renter_id)) AS retention      
FROM bookings AS b1
  LEFT JOIN bookings AS b2 ON
    b1.renter_id = b2.renter_id
    AND DATE_PART('year', b1.created_at) = DATE_PART(datetime('year', b2.created_at, '-1 year'))
GROUP BY  1
ORDER BY 2;

虽然它根本不起作用...... 我收到的错误消息是:提示:没有函数与给定的名称和参数类型匹配。您可能需要添加显式类型转换。

如果有一些建议会很棒。

Answer 1

直接的问题是 datetime() 函数，它在 PostgreSQL 中不存在。然后是别名和聚合的一些问题。这应该让你更接近你想去的地方:

SELECT 
  <b>b1.</b>renter_id,
  min(DATE_PART('year', <b>b1.</b>created_at)) AS first_rental_year,
  <strike>DATE_PART('year', created_ay) AS y1,</strike>            -- Needs an aggregate
  round(100<b>.</b> * count(distinct b2.renter_id) /     -- Use float, see 100.
    count(distinct b1.renter_id)) AS retention
FROM bookings AS b1
LEFT JOIN bookings AS b2
  ON  b1.renter_id = b2.renter_id
  AND extract(year, b1.created_at) + 1 = extract(year, b2.created_at)
GROUP BY 1
ORDER BY 2;

但是，您的同类群组功能的逻辑似乎存在缺陷。您应该将问题编辑为:

1. 更详细地说明您希望如何计算留存率
2. 添加表结构和一些示例数据
3. 显示预期结果。