按子查询分组并按平均值进行分类Bigquery

Question

我正在尝试按日期对记录进行分组，并针对该特定日期，获取称为延迟的列之一的平均值。我有以下标准SQL查询，该查询有效，但当我尝试获取表的平均值和分组依据时不起作用：

SELECT * 
From 
(
SELECT 'ADS-B Average Latency for Asia' as metric_name, 'Daily' as metric_period_type,
  DATE(timestamp) as metric_date, collection_type, UNIX_SECONDS(ingestion_time) - UNIX_SECONDS(timestamp) as latency
  FROM `analytics.aoi_table` a
  order by metric_date
) table 

这是表：

当我尝试使用下面的sql进行上表的平均值和分组时，出现以下错误：语法错误：预期为“）”，但关键字关键字GROUP在[10：3]

导致错误的SQL：

#standardSQL

SELECT * 
From 
(
SELECT 'ADS-B Average Latency for Asia' as metric_name,
  DATE(timestamp) as metric_date, collection_type, AVG(UNIX_SECONDS(ingestion_time) - UNIX_SECONDS(timestamp)) as latency
  FROM `ais-data-analysis.customer_analytics.itochu_aoi_table` a
  order by metric_date
  group by metric_date, collection_type
) table 

Answer 1

您的group by和order by乱序。另外，您不需要子查询：

select 'ADS-B Average Latency for Asia' as metric_name,
       DATE(timestamp) as metric_date, collection_type, AVG(UNIX_SECONDS(ingestion_time) - UNIX_SECONDS(timestamp)) as latency
from `ais-data-analysis.customer_analytics.itochu_aoi_table` a
group by metric_date, collection_type
order by metric_date

Answer 2

order by子句位于group by子句之后。另外，您也无需包装查询，只需执行以下操作即可：

SELECT 
    'ADS-B Average Latency for Asia' as metric_name,
    DATE(timestamp) as metric_date, 
    collection_type, 
    AVG(UNIX_SECONDS(ingestion_time) - UNIX_SECONDS(timestamp)) as latency
FROM `ais-data-analysis.customer_analytics.itochu_aoi_table` a
GROUP BY metric_date, collection_type
ORDER BY metric_date, collection_type