lag()函数的所有值都为null

Question

Select date, symbol, open, high, low, cts.close, volume, lag(cts.close) over (PARTITION by date, symbol) as prevclose, 
((cts.close - cts.open) / cts.close) * 100 as prcnt
from cts
where cts.close <> 0;

保持整列prevclose为null。

Answer 1

这很可能是因为您正在按作为时间组件的date列进行分区。每个分区都是唯一的，因此没有"prev"值。

没有CCD_ 3的CCD_ 2也是没有意义的。订单没有保证。

SELECT ..., lag(cts.close) over (PARTITION by date, symbol) as prevclose
FROM cts;

在我看来应该是

SELECT ..., lag(cts.close) over (PARTITION BY symbol ORDER BY date) as prevclose
FROM cts;