1

My question has been addressed previously but I can't seem to apply any solution to my query to make it work. Would very much appreciate some guidance.

My current query below returns this data set:

  |  Age      |   Count   |
     0-1 day      300
     2-3 days     6000
     3-4 days     100 

SELECT(CASE WHEN time_dtm > SYSDATE -1 THEN '0-1 day'
        WHEN time_dtm > SYSDATE -2 AND time_dtm < SYSDATE -1 THEN '1-2 days'
        WHEN time_dtm > SYSDATE -3 AND time_dtm < SYSDATE -2 THEN '2-3 days'
        WHEN time_dtm > SYSDATE -4 AND time_dtm < SYSDATE -3 THEN '3-4 days'
        WHEN time_dtm > SYSDATE -5 AND time_dtm < SYSDATE -4 THEN 'Closed'
        END) AS Age, 
    COUNT( * ) AS "Count" 
FROM table_1 
WHERE id IN (1,2,3)
GROUP BY (CASE WHEN time_dtm > SYSDATE -1 THEN '0-1 day'
        WHEN time_dtm > SYSDATE -2 AND time_dtm < SYSDATE -1 THEN '1-2 days'
        WHEN time_dtm > SYSDATE -3 AND time_dtm < SYSDATE -2 THEN '2-3 days'
        WHEN time_dtm > SYSDATE -4 AND time_dtm < SYSDATE -3 THEN '3-4 days'
        WHEN time_dtm > SYSDATE -5 AND time_dtm < SYSDATE -4 THEN 'Closed'
        END)
ORDER BY (CASE WHEN time_dtm > SYSDATE -1 THEN '0-1 day'
        WHEN time_dtm > SYSDATE -2 AND time_dtm < SYSDATE -1 THEN '1-2 days'
        WHEN time_dtm > SYSDATE -3 AND time_dtm < SYSDATE -2 THEN '2-3 days'
        WHEN time_dtm > SYSDATE -4 AND time_dtm < SYSDATE -3 THEN '3-4 days'
        WHEN time_dtm > SYSDATE -5 AND time_dtm < SYSDATE -4 THEN 'Closed'
        END)

However, I would like it to show the zero/null rows as zeros like this:

  |  Age      |   Count   |
     0-1 day      300
     1-2 days     0
     2-3 days     6000
     3-4 days     100
     Closed       0

I've read all sorts from the past couple of days re: NVL, COALESCE, FULL/LEFT/RIGHT OUTER JOIN, LEFT/RIGHT JOINS, UNION ALL etc none of which had CASE statements and tried to work around it myself BUT! You have to know when to stop and ask for directions.

Improve this question

1 Answer 1

Reset to default
3

First off, re-write your query. Use views or common table expression to avoid repeating yourself three times for your SELECT, GROUP BY, ORDER BY clauses. Your query becomes:

WITH data AS (
    SELECT(CASE WHEN time_dtm > SYSDATE -1 THEN '0-1 day'
                WHEN time_dtm > SYSDATE -2 AND 
                     time_dtm < SYSDATE -1 THEN '1-2 days'
                WHEN time_dtm > SYSDATE -3 AND 
                     time_dtm < SYSDATE -2 THEN '2-3 days'
                WHEN time_dtm > SYSDATE -4 AND 
                     time_dtm < SYSDATE -3 THEN '3-4 days'
                WHEN time_dtm > SYSDATE -5 AND 
                     time_dtm < SYSDATE -4 THEN 'Closed'
           END) AS Age 
    FROM table_1 
    WHERE id IN (1,2,3)
)
SELECT Age, COUNT(*)
FROM data
GROUP BY Age
ORDER BY Age

Then, in order to be sure that any of your desired groups will be available in the result, you have lots of options.

You could use UNION ALL:

WITH data AS (
    SELECT(CASE WHEN time_dtm > SYSDATE -1 THEN '0-1 day'
                WHEN time_dtm > SYSDATE -2 AND 
                     time_dtm < SYSDATE -1 THEN '1-2 days'
                WHEN time_dtm > SYSDATE -3 AND 
                     time_dtm < SYSDATE -2 THEN '2-3 days'
                WHEN time_dtm > SYSDATE -4 AND 
                     time_dtm < SYSDATE -3 THEN '3-4 days'
                WHEN time_dtm > SYSDATE -5 AND 
                     time_dtm < SYSDATE -4 THEN 'Closed'
           END) AS Age 
    FROM table_1 
    WHERE id IN (1,2,3)

    -- The below will add one record for every desired Age group
    UNION ALL
    SELECT '0-1 day'  FROM DUAL UNION ALL
    SELECT '1-2 days' FROM DUAL UNION ALL
    SELECT '2-3 days' FROM DUAL UNION ALL
    SELECT '3-4 days' FROM DUAL UNION ALL
    SELECT 'Closed'   FROM DUAL
)
SELECT Age, COUNT(*) - 1 -- Subtract the extra record again
FROM data
GROUP BY Age
ORDER BY Age

An entirely different solution would involve LEFT OUTER JOINs:

-- Groups is a dynamic table that contains the date ranges and their "Age" label
WITH groups AS (
    SELECT SYSDATE -1 lower, SYSDATE upper, '0-1 day'  Age FROM DUAL UNION ALL
    SELECT SYSDATE -2      , SYSDATE -1   , '1-2 days'     FROM DUAL UNION ALL
    SELECT SYSDATE -3      , SYSDATE -2   , '2-3 days'     FROM DUAL UNION ALL
    SELECT SYSDATE -4      , SYSDATE -3   , '3-4 days'     FROM DUAL UNION ALL
    SELECT SYSDATE -5      , SYSDATE -4   , 'Closed'       FROM DUAL
)
SELECT g.Age, NVL(SUM(t.counter), 0)
FROM groups g

-- LEFT OUTER JOINing "table_1" to "groups" will ensure that every group
-- appears at least once in the result
LEFT OUTER JOIN (
  SELECT 1 counter, t.* FROM table_1 t WHERE t.id IN (1,2,3)
) t
ON  t.time_dtm >= g.lower
AND t.time_dtm <  g.upper
GROUP BY g.Age
ORDER BY g.Age

In the second example, you could also do without a CTE and use a nested SELECT for the groups table. It is easy to see how the second example is simpler to evolve in the future, should your requirements change.

Improve this answer
Sign up to request clarification or add additional context in comments.

6 Comments

Thanks @Lukas, for answering my question. I tried the WITH DATA AS & UNION ALL but it did not recognise "Age" as an identifier. I also tried the dynamic groups & LEFT OUTER JOIN and it returned the same data set as per my original query (without zero rows). Is there anything else you would suggest? I'm using Toad for Oracle 11.5 if that if of relevance. ps. Dynamic groups is cool!
@ltran: I had a syntax error in my first query (forgot the FROM clause)... Also, in the second query, I have moved the t.id IN (1,2,3) predicate into a subselect. That was the reason why it didn't work. Let me know if this helps
Super awesome, they both work a treat! And thanks for the additional simplification of my query. I don't have enough points to vote yet but I will when they're accumulated @Lukas. Cheers.
Great to know. You can always "accept", without the necessary reputation to vote, I think
@ltran: Why -1? No records are added to table_1, the counts should correct...
|

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.