如何使用 LAG 函数进行 STDEV 计算？

Question

I'm running code like this:

SELECT ID, Date, Price,
    STDEV(Price) OVER (ORDER BY ID, Date ROWS BETWEEN 30 PRECEDING AND CURRENT ROW) As OneMonths,
    STDEV(Price) OVER (ORDER BY ID, Date ROWS BETWEEN 60 PRECEDING AND CURRENT ROW) As TwoMonths,
    STDEV(Price) OVER (ORDER BY ID, Date ROWS BETWEEN 90 PRECEDING AND CURRENT ROW) As ThreeMonths
FROM Price_Table

That gives me this result.

In the fiver first row I always have three nulls for the three variances. This makes sense. However, every time the ID changes, I must be getting the preceding ID's prices, because each time the ID changes, I would expect the standard deviation to get reset. So, the first line in orange should be null, I think, and the next one should be zero, because there is no change in price the second day. I tried wrapping the LAG function around the STDEV function and I got an error. I tried the opposite and also got an error.

Answer 1

If you want the value per id , then you need partition by :

SELECT ID, Date, Price,
       STDEV(Price) OVER (PARTITION BY ID ORDER BY Date ROWS BETWEEN 30 PRECEDING AND CURRENT ROW) As OneMonths,
       STDEV(Price) OVER (PARTITION BY ID ORDER BY Date ROWS BETWEEN 60 PRECEDING AND CURRENT ROW) As TwoMonths,
       STDEV(Price) OVER (PARTITION BY ID ORDER BY Date ROWS BETWEEN 90 PRECEDING AND CURRENT ROW) As ThreeMonths
FROM Price_Table;