Spread out x-axis points to smooth out line graph

Question

Hi guys,
I've been given some figures that I've displayed into a graph...However, there are some values that have been taken every 2 weeks and then there's like a 6 month gap. I'm wondering if it's possible to space out the dates relevent to duration in order to make the learn smoother. For example:
http://i29.tinypic.com/rcv8rr.png
If there was like 20 plots on the graph then the closer dates might be spaced out 0.2cm while the larger ones may be like 0.8cm etc.
Mike

Answer 1

Hi Mike,
OK - I have not found a simple function or chart setting that will "fill in the gaps" in the data. If anyone knows of one, let me know!
So, here's some more SQL for you:
Firstly, assuming that data needs to be shown weekly:
SELECT
NULL LINK,
Y.NEW_DATE LABEL,
CASE WHEN Y.NEW_DATE = Y.LAST_HIREDATE THEN Y.EMPNO
    ELSE ROUND((Y.LAST_EMPNO + (((Y.NEW_DATE - Y.LAST_HIREDATE) / (Y.NEXT_HIREDATE - Y.LAST_HIREDATE))) * (Y.NEXT_EMPNO - Y.LAST_EMPNO)),0)
  END VALUE
FROM (
SELECT X.NEW_DATE, E.EMPNO,
(SELECT TRUNC(MAX(E2.HIREDATE),'DAY') FROM EMP E2 WHERE TRUNC(E2.HIREDATE,'DAY') <= X.NEW_DATE) LAST_HIREDATE,
(SELECT MAX(E3.EMPNO) FROM EMP E3 WHERE TRUNC(E3.HIREDATE,'DAY') = (SELECT TRUNC(MAX(E4.HIREDATE),'DAY') FROM EMP E4 WHERE TRUNC(E4.HIREDATE,'DAY') <= X.NEW_DATE)) LAST_EMPNO,
(SELECT TRUNC(MIN(E5.HIREDATE),'DAY') FROM EMP E5 WHERE TRUNC(E5.HIREDATE,'DAY') >= X.NEW_DATE) NEXT_HIREDATE,
(SELECT MIN(E6.EMPNO) FROM EMP E6 WHERE TRUNC(E6.HIREDATE,'DAY') = (SELECT TRUNC(MIN(E7.HIREDATE),'DAY') FROM EMP E7 WHERE TRUNC(E7.HIREDATE,'DAY') >= X.NEW_DATE)) NEXT_EMPNO
FROM (
SELECT (SELECT TRUNC(MIN(HIREDATE),'DAY') FROM EMP) + ((LEVEL - 1) * 7) NEW_DATE
FROM DUAL
CONNECT BY (SELECT TRUNC(MIN(HIREDATE),'DAY') FROM EMP) + ((LEVEL - 1) * 7) < (SELECT MAX(HIREDATE) FROM EMP)
) X
LEFT OUTER JOIN EMP E ON TRUNC(E.HIREDATE,'DAY') = X.NEW_DATE
ORDER BY X.NEW_DATE
) YOr, monthly:
SELECT
NULL LINK,
Y.NEW_DATE LABEL,
CASE WHEN Y.NEW_DATE = Y.LAST_HIREDATE THEN Y.EMPNO
    ELSE ROUND((Y.LAST_EMPNO + (((Y.NEW_DATE - Y.LAST_HIREDATE) / (Y.NEXT_HIREDATE - Y.LAST_HIREDATE))) * (Y.NEXT_EMPNO - Y.LAST_EMPNO)),0)
  END VALUE
FROM (
SELECT X.NEW_DATE, E.EMPNO,
(SELECT TRUNC(MAX(E2.HIREDATE),'MONTH') FROM EMP E2 WHERE TRUNC(E2.HIREDATE,'MONTH') <= X.NEW_DATE) LAST_HIREDATE,
(SELECT MAX(E3.EMPNO) FROM EMP E3 WHERE TRUNC(E3.HIREDATE,'MONTH') = (SELECT TRUNC(MAX(E4.HIREDATE),'MONTH') FROM EMP E4 WHERE TRUNC(E4.HIREDATE,'MONTH') <= X.NEW_DATE)) LAST_EMPNO,
(SELECT TRUNC(MIN(E5.HIREDATE),'MONTH') FROM EMP E5 WHERE TRUNC(E5.HIREDATE,'MONTH') >= X.NEW_DATE) NEXT_HIREDATE,
(SELECT MIN(E6.EMPNO) FROM EMP E6 WHERE TRUNC(E6.HIREDATE,'MONTH') = (SELECT TRUNC(MIN(E7.HIREDATE),'MONTH') FROM EMP E7 WHERE TRUNC(E7.HIREDATE,'MONTH') >= X.NEW_DATE)) NEXT_EMPNO
FROM (
SELECT ADD_MONTHS((SELECT TRUNC(MIN(HIREDATE),'MONTH') FROM EMP), LEVEL - 1) NEW_DATE
FROM DUAL
CONNECT BY ADD_MONTHS((SELECT TRUNC(MIN(HIREDATE),'MONTH') FROM EMP), LEVEL - 1) < (SELECT MAX(HIREDATE) FROM EMP)
) X
LEFT OUTER JOIN EMP E ON TRUNC(E.HIREDATE,'MONTH') = X.NEW_DATE
ORDER BY X.NEW_DATE
) YI appreciate that these are a tad complicated but they are based on a simple principle:
This will create a table of dates for the first day in each month for the next 6 months:
SELECT ADD_MONTHS(TRUNC(SYSDATE,'MONTH'), LEVEL-1) MONTH_BEGIN
FROM DUAL
CONNECT BY LEVEL <= 6Similar, but for every month from now until the end of next year:
SELECT ADD_MONTHS(TRUNC(SYSDATE,'MONTH'), LEVEL-1) MONTH_BEGIN
FROM DUAL
CONNECT BY ADD_MONTHS(TRUNC(SYSDATE,'MONTH'), LEVEL-1) < TO_DATE('31/12/2010','DD/MM/YYYY')These are two examples of how to create virtual records using a loop. LEVEL indicates the current iteration number and CONNECT BY acts like a WHILE statement
Using such records, we can link to existing data using a LEFT OUTER JOIN. This way, we keep all the dates and populate ones that have actual values. For all records, I'm noting the last actual date/value that exists before it and the first actual date/value that exists after it. This gives me a complete set of data. Leaving it at that stage, it pretty much where my last post's statement was.
I have now extended this to perform a calculation based on how far between two actual values the current record is - based on the dates. This then gives me a percentage that I can add to the last know value to get a derived value that will fill the gap.
I have used the "Monthly" sql version here: [http://apex.oracle.com/pls/otn/f?p=9568:11] Which will, hopefully, make things a bit clearer as you can see the actual values in a report on the right but the filled in values in the chart
Andy