Question

我在BigQuery中有一个表格，其中包含每天发生某些交易时的净库存量以及具有价格历史记录的表格，如下所示：

WITH `trans` AS (
SELECT DATE '2018-10-02' trans_date, 10.0 quantity UNION ALL
SELECT DATE '2018-10-03', 5.0 UNION ALL
SELECT DATE '2018-10-05', 11.0 
),
`prices` AS (
SELECT DATE '2018-10-01' price_date, 1.0 price UNION ALL
SELECT DATE '2018-10-02', 2.0 UNION ALL
SELECT DATE '2018-10-03', 3.0 UNION ALL
SELECT DATE '2018-10-04', 4.0 UNION ALL
SELECT DATE '2018-10-05', 5.0 UNION ALL
SELECT DATE '2018-10-06', 6.0 UNION ALL
SELECT DATE '2018-10-07', 7.0 
)
SELECT 
price_date,
quantity, 
price
FROM (
SELECT price_date, quantity, price, trans_date FROM `trans`
RIGHT JOIN `prices`
ON trans.trans_date = prices.price_date
ORDER BY price_date
)

我不知道如何填写该日期没有交易的最后一个已知数量，以得到如下结果：

price_date    quantity  price
2018-10-01    0.0    1.0
2018-10-02    10.0   2.0
2018-10-03    5.0    3.0
2018-10-04    5.0    4.0
2018-10-05    11.0   5.0
2018-10-06    11.0   6.0
2018-10-07    11.0   7.0

下一步将是随着时间的推移计算值。对于任何示例/建议，我将不胜感激。

编辑：我添加了股票代号，以匹配更真实的示例

WITH `trans` AS (
SELECT DATE '2018-10-02' trans_date, 10.0 quantity, 'TX' symbol UNION ALL
SELECT DATE '2018-10-03' trans_date, 5.0 quantity, 'TX' UNION ALL
SELECT DATE '2018-10-05', 11.0, 'AX' 
),
`prices` AS (
 SELECT DATE '2018-10-01' price_date, 1.0 price, 'TX' symbol UNION ALL
  SELECT DATE '2018-10-02', 2.0, 'TX' UNION ALL
  SELECT DATE '2018-10-03', 3.0, 'TX' UNION ALL
  SELECT DATE '2018-10-04', 4.0, 'TX' UNION ALL
  SELECT DATE '2018-10-05', 5.0, 'TX' UNION ALL
  SELECT DATE '2018-10-06', 6.0, 'TX' UNION ALL
  SELECT DATE '2018-10-07', 7.0, 'TX' UNION ALL
  SELECT DATE '2018-10-08', 8.0, 'AX' UNION ALL
  SELECT DATE '2018-10-09', 9.0, 'TX' UNION ALL
  SELECT DATE '2018-10-10', 10.0, 'AX' UNION ALL
  SELECT DATE '2018-10-11', 11.0, 'TX' UNION ALL
  SELECT DATE '2018-10-12', 12.0, 'TX' 
)

SELECT
  price_date, 
  t.symbol AS symbol,
  IFNULL(
    ARRAY_AGG(
      IF(p.price_date >= t.trans_date AND p.symbol = t.symbol, quantity, NULL) 
      IGNORE NULLS ORDER BY trans_date DESC LIMIT 1
      )[OFFSET(0)],
  -1234567890) quantity,
  price
FROM `prices` p
CROSS JOIN `trans` t
GROUP BY price_date, price, symbol
HAVING quantity != -1234567890
ORDER BY price_date

Answer 1

您可以这样做：

select p.price_date,
       coalesce(t.quantity, lag(t.quantity ignore nulls) over (order by p.price_date)) as quantity
       p.price
from prices p join
     trans t
     on t.trans_date = p.price_date;

编辑：

是的。 IGNORE NULL在某些情况下有效，但在其他情况下则无效。数组通常是BigQuery中的解决方案：

select price_date,
       (select quantity
        from unnest(quantities) quantity with OFFSET n
        where quantity is not null
        order by n desc
        LIMIT 1
       ) as quantity,
       price
from (select p.price_date,
             array_agg(t.quantity) over (order by p.price_date) as quantities,
             p.price
      from prices p LEFT join
           trans t
           on t.trans_date = p.price_date
    ) pp;

Answer 2

以下是用于BigQuery标准SQL

#standardSQL
WITH `trans` AS (
  SELECT DATE '2018-10-02' trans_date, 10.0 quantity UNION ALL
  SELECT DATE '2018-10-03', 5.0 UNION ALL
  SELECT DATE '2018-10-05', 11.0 
),
`prices` AS (
  SELECT DATE '2018-10-01' price_date, 1.0 price UNION ALL
  SELECT DATE '2018-10-02', 2.0 UNION ALL
  SELECT DATE '2018-10-03', 3.0 UNION ALL
  SELECT DATE '2018-10-04', 4.0 UNION ALL
  SELECT DATE '2018-10-05', 5.0 UNION ALL
  SELECT DATE '2018-10-06', 6.0 UNION ALL
  SELECT DATE '2018-10-07', 7.0 
)
SELECT 
  price_date, 
  IFNULL(
    ARRAY_AGG(
      IF(p.price_date >= t.trans_date, quantity, NULL) 
      IGNORE NULLS ORDER BY trans_date DESC LIMIT 1
      )[OFFSET(0)],
  0) quantity,
  price
FROM `prices` p
CROSS JOIN `trans` t
GROUP BY price_date, price
-- ORDER BY price_date

填写BigQuery中联接表的缺失值

2 个答案: