How to group data results by month - oracle

Question

I have a script that I wrote that will tell me what our "On Time Delivery" or "OTD" is based on the current month. Essentially it tells us instantly if we have shipped our materials before or on our promise date.
What I am trying to do now is create a new script that will tie this into our GLPERIODS table so that I can show what the "OTD" is for each month of the year at a glance. When I run this new I get the correct period (Jan-Dec) but my OTD calculation is not grouped by period. Instead they are all the same and I think it's an average of all the data in our system.
I will post the script here as well as the results. (edit: I guess I can't post the results as it won't let me post an image) Any help would be greatly appreciated. Thank you and BTW, I am not a programmer, just someone who works on this stuff occasionally to get the data we are after.
Select "month",
OTD
From (Select Round((1 - ("Late" / "All")), 3) As OTD
From (Select Round(Count(V_RELEASES_COMB.ACTUAL_SHIPDATE), 2) As "Late"
From V_RELEASES_COMB,
GLPERIODS
Where To_Char(V_RELEASES_COMB.ACTUAL_SHIPDATE, 'MM/YYYY') =
To_Char(GLPERIODS.START_DATE, 'MM/YYYY') And
V_RELEASES_COMB.ACTUAL_SHIPDATE > V_RELEASES_COMB.PROMISE_DATE And
V_RELEASES_COMB.ITEMNO Is Not Null And V_RELEASES_COMB.CUMM_SHIPPED > 0),
(Select Case
When Round(Count(V_RELEASES_COMB.ACTUAL_SHIPDATE), 2) = 0 Then 1
Else Round(Count(V_RELEASES_COMB.ACTUAL_SHIPDATE), 2) End As "All"
From V_RELEASES_COMB,
GLPERIODS
Where To_Char(V_RELEASES_COMB.ACTUAL_SHIPDATE, 'MM/YYYY') =
To_Char(GLPERIODS.START_DATE, 'MM/YYYY') And V_RELEASES_COMB.ITEMNO Is Not
Null And V_RELEASES_COMB.CUMM_SHIPPED > 0)),
(Select To_Char(GLPERIODS.START_DATE, 'MONTH') As "month"
From GLPERIODS
Where Extract(Year From GLPERIODS.START_DATE) = Extract(Year From SysDate))
Ok, thank you for the help! Here is the edited script with your input added:
Select "month",
OTD
From (Select trunc(g.start_date, 'Month') mth,
round ( 1 - ( count(
case when v.ACTUAL_SHIPDATE > v.PROMISE_DATE And v.ITEMNO Is Not Null
And v.CUMM_SHIPPED > 0 then V.ACTUAL_SHIPDATE end)
/ greatest(Count(
case when v.ITEMNO Is Not Null And v.CUMM_SHIPPED > 0 then V.ACTUAL_SHIPDATE end
), 1) ) , 3) OTD
from glperiods g
left join v_releases_comb v
on trunc(v.ACTUAL_SHIPDATE, 'Month') = trunc(g.start_date, 'Month')
group by trunc(g.start_date, 'Month')),
(Select To_Char(GLPERIODS.START_DATE, 'MONTH') As "month"
From GLPERIODS
Where Extract(Year From GLPERIODS.START_DATE) = Extract(Year From SysDate))
The results look promising but here is what is happening:
Month OTD
January .956
January .875
January .359
January 1
January 1
January 1
February .978
February .562
February .875
February 1
February 1
etc.
this continues for all months.
The above is in a table format but I have no idea how to post a table on this site. I wish I could just upload a pic but it won't let me.

Answer 1

This query should do the job:
SQLFiddle
select to_char(mth, 'Month') "Month", OTD
from (Select trunc(g.start_date, 'Month') mth,
round ( 1 - ( count(
case when v.ACTUAL_SHIPDATE > v.PROMISE_DATE And v.ITEMNO Is Not Null
And v.CUMM_SHIPPED > 0 then V.ACTUAL_SHIPDATE end)
/ greatest(Count(
case when v.ITEMNO Is Not Null And v.CUMM_SHIPPED > 0 then V.ACTUAL_SHIPDATE end
), 1) ) , 3) OTD
from glperiods g
left join v_releases_comb v
on trunc(v.ACTUAL_SHIPDATE, 'Month') = trunc(g.start_date, 'Month')
Where Extract(Year From g.START_DATE) = Extract(Year From SysDate)
group by trunc(g.start_date, 'Month'))
order by mth
Without few examples it's hard to say for sure that it's OK, it would be good to verify with something.
So if something is wrong please edit your post, add some example rows and explain.
Edit:
I think that creating the view may be more comfortable for you:
create or replace view v_monthly_sums as
Select trunc(g.start_date, 'Month') mth,
count(case when v.ACTUAL_SHIPDATE > v.PROMISE_DATE And v.ITEMNO Is Not Null
And v.CUMM_SHIPPED > 0 then V.ACTUAL_SHIPDATE end) d1,
Count(
case when v.ITEMNO Is Not Null And v.CUMM_SHIPPED > 0 then V.ACTUAL_SHIPDATE end
) d2
from glperiods g
left join v_releases_comb v
on trunc(v.ACTUAL_SHIPDATE, 'Month') = trunc(g.start_date, 'Month')
group by trunc(g.start_date, 'Month')
From this view you can select all values (select * from v_monthly_sums), to see what it contains, do any experiments,
but for your main question you can use something like:
select to_char(mth, 'Month') "Month",
case
when d2 = 0 then null
else round(1-d1/d2, 3)
end otd
from v_monthly_sums
where extract(Year From mth) = 2015 -- you can add here: AND D2<>0 to eliminate null rows
order by mth
GLPERIODS - I suspect that this table is not needed at all, not needed in our query/view for sure,
maybe you created it only to store dates for months which interests you.
But maybe I'm wrong and it is used somewhere else. It can by replaced in our view with simple hierarchical subquery,
but for now let's stay with current solution. Please insert one date for each month, I did not make any protections against more dates.
Why did not your query work? - you joined one number, let's say 0,567, produced by two other complicated queries with every day from table GLPERIODS. This is the reason why you got 12 months with the same, invaluable data.