Google bigquery BigQuery标准在分组时获取第一个非空值
我有一张这样的桌子:Google bigquery BigQuery标准在分组时获取第一个非空值,google-bigquery,coalesce,Google Bigquery,Coalesce,我有一张这样的桌子: CUSTOMERS_ID DATE_SALES DIMENSION MARIO1 20200201 NULL MARIO1 20200113 Spain MARIO2 20200131 NULL MARIO3 20200101 France MARIO3 20191231 Spain 我需要按客户编号和日期销售描述字段进行订购。然后我想按CUSTOMERS\u ID字段
CUSTOMERS_ID DATE_SALES DIMENSION
MARIO1 20200201 NULL
MARIO1 20200113 Spain
MARIO2 20200131 NULL
MARIO3 20200101 France
MARIO3 20191231 Spain
我需要按客户编号和日期销售描述字段进行订购。然后我想按CUSTOMERS\u ID字段分组,并获取维度字段的第一个非空值。
输出表将是:
CUSTOMERS_ID DIMENSION
MARIO1 Spain
MARIO2 NULL
MARIO3 France
有什么想法吗?我尝试过合并函数,第一个值,但没有得到我预期的结果
提前谢谢 我们可以在这里使用行号技巧:
WITH cte AS (
SELECT CUSTOMERS_ID,
ROW_NUMBER() OVER (PARTITION BY CUSTOMERS_ID
ORDER BY -1.0*UNIX_SECONDS(DATE_SALES) DESC) rn
FROM yourTable
)
SELECT CUSTOMERS_ID, DIMENSION
FROM cte
WHERE rn = 1
ORDER BY CUSTOMERS_ID;
逻辑是将行数按从历元开始的负秒数降序。这会将最近的销售放在第一位,也会将NULL放在最后,因此如果没有非NULL维度数据可用,则NULL值只会收到行号1。我们可以在此处使用行号技巧:
WITH cte AS (
SELECT CUSTOMERS_ID,
ROW_NUMBER() OVER (PARTITION BY CUSTOMERS_ID
ORDER BY -1.0*UNIX_SECONDS(DATE_SALES) DESC) rn
FROM yourTable
)
SELECT CUSTOMERS_ID, DIMENSION
FROM cte
WHERE rn = 1
ORDER BY CUSTOMERS_ID;
逻辑是将行数按从历元开始的负秒数降序。这会将最近的销售放在第一位,也会将NULL放在最后,因此,如果没有非NULL维度数据可用,则NULL值只会收到第1行。您可以按客户id分组,并通过忽略NULL使用ARRAY_AGG,还可以在该字段中按日期订购。 限制1将通过使用更少的RAM存储来提高效率。 然后,OFFSET0将使它成为一个不需要赋值的字段,因此您可以轻松地使用该字段
WITH
raw_data AS
(
SELECT 'MARIO1' CUSTOMERS_ID, 20200201 DATE_SALES, NULL as DIMENSION UNION ALL
SELECT 'MARIO1' CUSTOMERS_ID, 20200113 DATE_SALES, 'Spain' as DIMENSION UNION ALL
SELECT 'MARIO2' CUSTOMERS_ID, 20200131 DATE_SALES, NULL as DIMENSION UNION ALL
SELECT 'MARIO3' CUSTOMERS_ID, 20200101 DATE_SALES, 'France' as DIMENSION UNION ALL
SELECT 'MARIO3' CUSTOMERS_ID, 20191231 DATE_SALES, 'Spain' as DIMENSION
)
SELECT CUSTOMERS_ID, ARRAY_AGG(DIMENSION IGNORE NULLS ORDER BY DATE_SALES DESC LIMIT 1)[OFFSET(0)] as DIMENSION
FROM raw_data
GROUP BY 1
您可以按客户id分组,并通过忽略空值使用数组_AGG,还可以在该字段中按日期订购。 限制1将通过使用更少的RAM存储来提高效率。 然后,OFFSET0将使它成为一个不需要赋值的字段,因此您可以轻松地使用该字段
WITH
raw_data AS
(
SELECT 'MARIO1' CUSTOMERS_ID, 20200201 DATE_SALES, NULL as DIMENSION UNION ALL
SELECT 'MARIO1' CUSTOMERS_ID, 20200113 DATE_SALES, 'Spain' as DIMENSION UNION ALL
SELECT 'MARIO2' CUSTOMERS_ID, 20200131 DATE_SALES, NULL as DIMENSION UNION ALL
SELECT 'MARIO3' CUSTOMERS_ID, 20200101 DATE_SALES, 'France' as DIMENSION UNION ALL
SELECT 'MARIO3' CUSTOMERS_ID, 20191231 DATE_SALES, 'Spain' as DIMENSION
)
SELECT CUSTOMERS_ID, ARRAY_AGG(DIMENSION IGNORE NULLS ORDER BY DATE_SALES DESC LIMIT 1)[OFFSET(0)] as DIMENSION
FROM raw_data
GROUP BY 1
下面是BigQuery标准SQL
#standardSQL
SELECT AS VALUE ARRAY_AGG(t ORDER BY IF(DIMENSION IS NULL, NULL, DATE_SALES) DESC LIMIT 1)[OFFSET(0)]
FROM `project.dataset.table` t
GROUP BY CUSTOMERS_ID
如果要应用于您问题中的样本数据-结果为
Row CUSTOMERS_ID DATE_SALES DIMENSION
1 MARIO1 20200113 Spain
2 MARIO2 20200131 null
3 MARIO3 20200101 France
下面是BigQuery标准SQL
#standardSQL
SELECT AS VALUE ARRAY_AGG(t ORDER BY IF(DIMENSION IS NULL, NULL, DATE_SALES) DESC LIMIT 1)[OFFSET(0)]
FROM `project.dataset.table` t
GROUP BY CUSTOMERS_ID
如果要应用于您问题中的样本数据-结果为
Row CUSTOMERS_ID DATE_SALES DIMENSION
1 MARIO1 20200113 Spain
2 MARIO2 20200131 null
3 MARIO3 20200101 France
ARRAY_AGG和OFFSET是我的新好朋友。谢谢你,米哈伊尔!ARRAY_AGG和OFFSET是我的新好朋友。谢谢你,米哈伊尔!