Question

I currently have data stored as follows:

PERSON      DATE        RATE
----------------------------
John Smith  1/4/2012    1.2
John Smith  8/6/2012    1.7
John Smith  8/13/2012   1.9
John Smith  8/20/2012   2
John Smith  9/10/2012   1.8
John Smith  10/1/2012   3

I’m trying to output a rate for each week (ending Sunday) of the year for each person. Where the rate doesn’t exist for a given week, the previous week is used i.e.:

PERSON      WEEK       RATE
----------------------------
John Smith  1/8/2012    1.2
John Smith  1/15/2012   1.2
John Smith  1/22/2012   1.2
John Smith  1/29/2012   1.2

etc

I can build a table of for date and week combinations so I can determine the week. How can I duplicate the rate though? An outer join or something similar?

Answer 1

There’s a lot I don’t know about your data, but here’s a suggestion. The first subquery selects all distinct persons, the second matches each of these to each week (supposing you store weeks as date and 00:00 am Sunday morning). This supposes that Person is unique, but I hope you have a primary key that can be used to determine who’s who.

SELECT 
    X.Person,
    W.Date,
    X.Rate
FROM
(
    SELECT
        Person
    FROM
        Ratings
    GROUP BY
        Person      
) P

CROSS JOIN
(
    SELECT
         Date AS DateEnd    
    FROM
        Weeks
) W

CROSS APPLY
(   -- Last Rate before week end
    SELECT TOP 1 
        Rate
    FROM 
        Ratings
    WHERE 
        Person = P.Person AND
        Date < Dateadd(DAY,1,W.DateEnd)
    ORDER BY 
        DATE DESC
) X