需要按期间（年，月）获取不同的累计计数

Question

我已经设法获得按月计算的累计非重复计数，但在尝试按年计算时却无法获得。我尝试了分区函数，但我不断收到各种错误，这就是我到目前为止所得到的。你能给我一些指导吗，我是新手。

[SQL Fiddle][1]

PostgreSQL 9.3 架构设置:

create table table2 (id serial, fecha_captura date, mn integer, VALUACION1 integer);

    insert into table2 (fecha_captura, mn, VALUACION1)
    values
    ('1996-02-01',2,12345),
    ('1996-02-01',2,12345),
    ('1996-02-01',2,12344),
    ('1996-02-01',2,12343),
    ('1996-03-01',3,12345),
    ('1996-03-01',3,12346),
    ('1996-03-01',3,12347),
    ('1996-03-01',3,12348),
    ('1996-04-01',4,12349),
    ('1996-04-01',4,12350),
    ('1996-04-01',4,12351),
    ('1996-04-01',4,12352),
    ('1997-03-01',4,12353),
    ('1997-03-01',4,12354),
    ('1997-03-01',4,12345),
    ('1997-03-01',4,12352),
    ('1997-04-01',4,12353),
    ('1997-04-01',4,12354),
    ('1997-04-01',4,12345),
    ('1997-04-01',4,12355);

查询 1:

SELECT Year1,mn,
   COUNT(DISTINCT(VALUACION1)) AS Val_Dist_Mes
  ,SUM(flag) AS Val_Dist_Mes_Nuevas
  ,SUM(SUM(flag)) 
   OVER (partition by Year1 ORDER BY mn,Year1
         ROWS UNBOUNDED PRECEDING) AS Val_Dist_Acum 
FROM
 (
   SELECT
      VALUACION1
     ,EXTRACT(MONTH FROM fecha_captura) AS mn
     ,EXTRACT(YEAR FROM fecha_captura) AS Year1
     ,CASE
         WHEN EXTRACT(MONTH FROM fecha_captura) 
              = MIN (EXTRACT(MONTH FROM fecha_captura))
              OVER (PARTITION BY VALUACION1) 
         THEN 1 
         ELSE 0 
      END AS flag
   FROM table2
   GROUP BY fecha_captura,VALUACION1
 ) AS dt
GROUP BY mn,Year1

[结果][2]:

| year1 | mn | val_dist_mes | val_dist_mes_nuevas | val_dist_acum |
|-------|----|--------------|---------------------|---------------|
|  1996 |  2 |            3 |                   3 |             3 |
|  1996 |  3 |            4 |                   3 |             6 |
|  1996 |  4 |            4 |                   3 |             9 |
|  1997 |  3 |            4 |                   3 |             3 |
|  1997 |  4 |            4 |                   1 |             4 |

[预期结果][2]:

| year1 | mn | val_dist_mes | val_dist_mes_nuevas | val_dist_acum  |
|-------|----|--------------|---------------------|----------------|
|  1996 |  2 |            3 |                   3 |             3  |
|  1996 |  3 |            4 |                   3 |             6  |
|  1996 |  4 |            4 |                   4 |             10 |
|  1997 |  3 |            4 |                   4 |             4  |
|  1997 |  4 |            4 |                   1 |             5  |

Answer 1

在 group by 和 partition by 中添加 Year 应该可以解决问题

SELECT AÑO, 
       mn, 
       Count(DISTINCT( valuacion1 ))             AS Val_Dist_Mes, 
       Sum(flag)                                 AS Val_Dist_Mes_Nuevas, 
       Sum(Sum(flag)) 
         OVER ( 
           partition BY AÑO 
           ORDER BY mn rows UNBOUNDED PRECEDING) AS Val_Dist_Acum 
FROM   (SELECT valuacion1, 
               Extract(month FROM fecha_captura) AS mn, 
               Extract(year FROM fecha_captura)  AS AÑO, 
               CASE -- find the first year when a customer placed an order 
                 WHEN Extract(month FROM fecha_captura) = Min ( 
                      Extract(month FROM fecha_captura)) 
                      OVER ( 
                        partition BY valuacion1, Extract(year 
                      FROM fecha_captura)) THEN 1 
                 ELSE 0 
               END                               AS flag 
        FROM   table2 
        GROUP  BY fecha_captura, 
                  valuacion1) AS dt 
GROUP  BY mn, 
          AÑO

SQL FIDDLE DEMO

需要按期间（年，月）获取不同的累计计数

Need to get cumulative count distinct by periods (year,month)

sql

postgresql

cumulative-frequency