Question

在某些情况下，我想要一些列。

列是： point_id，weight（行当前度指数0到7,0是最新行），localdate（YYYYmmdd），tmin（最低温度），tmax（最高温度）和precip_amount（降水量mm）。

条件是：
（localdate＆gt; = 20151201 AND localdate＆lt; = 20160104）AND （tmin <0或tmax <0）AND precip_amount＆gt; 40

我的目标是根据体重获取最新的行。

我编写了一个工作正常的sql（经过了2.7秒，处理了64.4 MB）。

但是，有没有办法优化我的SQL以便更快地运行？

我的sql：

select a.point_id as point_id , a.weight as min_weight,a.localdate as local_date, a.tmin as temp_min, a.tmax as temp_max, a.precip_amount as precipitation
from table1 a
join (select point_id, min(weight) as min_weight
            from 
            (select point_id, localdate, tmin, tmax, precip_amount,weight
            from table1
            where (localdate >= 20151201 and localdate <= 20160104)  and (tmin < 0 or tmax < 0) and precip_amount > 40
            order by weight)
            group by point_id) b
on a.point_id = b.point_id and a.weight = b.min_weight
where (a.localdate >= 20151201 and a.localdate <= 20160104)  and (a.tmin < 0 OR a.tmax < 0) and a.precip_amount > 40
order by a.weight, a.localdate

Answer 1

尝试此查询。我已经将关于最小权重的计算移动到子查询。

SELECT a.point_id as point_id, 
        (SELECT MIN(A2.weight) FROM Table A2
         WHERE A2.point_id = A.point_id
               AND A2.localdate >= 20151201 
               AND A2.localdate <= 20160104  
               AND (A2.tmin < 0 or A2.tmax < 0) 
               AND A2.precip_amount > 40) AS min_weight,
        A.localdate as local_date, 
        A.tmin as temp_min, 
        A.tmax as temp_max, 
        A.precip_amount as precipitation
FROM Table AS A
WHERE  A.localdate >= 20151201
       AND A.localdate <= 20160104 
       AND (A.tmin < 0 OR A.tmax < 0) 
       AND A.precip_amount > 40
ORDER BY A.weight, A.localdate

您也可以尝试使用窗口功能。例如

SELECT T.* FROM
       (SELECT A.point_id,
               MIN(A.weight) OVER(PARTITION BY A.point_id) AS min_weight,
               A.localdate AS local_date,
               A.tmin AS temp_min,
               A.tmax AS temp_max,
               A.precip_amount as precipitation
        FROM Table A
        WHERE A.localdate >= 20151201
                   AND A.localdate <= 20160104 
                   AND (A.tmin < 0 OR A.tmax < 0) 
                   AND A.precip_amount > 40) AS T
    ORDER BY T.min_weight, 
             T.localdate

Answer 2

select * 
from 
( select a.point_id as point_id , a.weight as min_weight,a.localdate as local_date, a.tmin as temp_min, a.tmax as temp_max, a.precip_amount as precipitation  
       , row_number() over (partition by point_id order by weight asc) as rn
    from table1 a
   where (a.localdate >= 20151201 and a.localdate <= 20160104)   
     and (a.tmin < 0 OR a.tmax < 0) 
     and a.precip_amount > 40
) tt
where tt.rn = 1
order by tt.weight, tt.localdate

Answer 3

尝试使用窗口功能：

select point_id, a.weight as min_weight, local_date, temp_min, temp_max,
       precip_amount
from (select a.point_id as point_id, a.weight,
             a.localdate as local_date, a.tmin as temp_min, a.tmax as temp_max,
             a.precip_amount as precipitation,
             min(weight) over (partition by point_id) as min_weight
     from table1 a
     where (a.localdate >= 20151201 and a.localdate <= 20160104) and
           (a.tmin < 0 OR a.tmax < 0) and a.precip_amount > 40
    ) a
where a.weight = w.min_weight
order by a.weight, a.localdate;

有没有办法优化我的bigquery sql以便更快地运行？

3 个答案: