Do you wonder if your IT business is truly profitable or if you should raise your prices? Learn how to calculate your overhead burden using our free interactive tool and use it to determine the right price for your IT services. Start calculating Now!
SQL Margin Calculation gives divide by zero errors
I have a table showing Revenue (R), Cost (C) & Margin (M) values by Product Category for each period 01 > 12 like in the attached
I have substituted the NULL values with 0 using CASE WHEN SUM(C03) IS NULL THEN 0 ELSE SUM(C03) END
In each period margin calc I use CASE to avoid the Divide by Zero errors when the Revenue values are 0 like
CAST (CASE WHEN Sum([R01])= 0 OR Sum([R01]) IS NULL OR Sum([C01]) = 0 OR Sum([C01]) IS NULL THEN 0 ELSE
((Sum([R01]) - Sum([C01]))/(Sum([R01])))*
However I cannot see how is best to create the margin values using the totals.
What I need is ((RTotal-CTotal)/RTotal)*1
Driving me crazy, thanks.
I have substituted the NULL values with 0 using CASE WHEN SUM(C03) IS NULL THEN 0 ELSE SUM(C03) END
In each period margin calc I use CASE to avoid the Divide by Zero errors when the Revenue values are 0 like
CAST (CASE WHEN Sum([R01])= 0 OR Sum([R01]) IS NULL OR Sum([C01]) = 0 OR Sum([C01]) IS NULL THEN 0 ELSE
((Sum([R01]) - Sum([C01]))/(Sum([R01])))*
However I cannot see how is best to create the margin values using the totals.
What I need is ((RTotal-CTotal)/RTotal)*1
Driving me crazy, thanks.
Period R01 C01 M01% R02 C02 M02% R03 C03 M03% R04 C04 M04% ....... R12 C12 M12% RTotal CTotal MTotal%
Product A 50 20 60 100 20 80 100 0 100 0 0 0 0 0 0 250 40 ???????
Do more with
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
Hi, could you provide 3 things?
1. the existing query in full
2. sample data (not the query result, the source data)
3. expected result
It appears you are taking normalized data and pivoting this into 36 columns, and you also want a grand total set of columns. (? is that correct ?)
The most efficient way of doing this may not be trying to reuse the 36 column summary.
--
please note I edited your question to display the attached data in a code block, these are better as we don't have to look at multiple sources to read a question. :-)
1. the existing query in full
2. sample data (not the query result, the source data)
3. expected result
It appears you are taking normalized data and pivoting this into 36 columns, and you also want a grand total set of columns. (? is that correct ?)
The most efficient way of doing this may not be trying to reuse the 36 column summary.
--
please note I edited your question to display the attached data in a code block, these are better as we don't have to look at multiple sources to read a question. :-)
Your assumption is correct.
The data, query and expected result are now attached.
Thanks.
Data---Expected-Result.xlsx
The data, query and expected result are now attached.
Thanks.
DECLARE @FromDate DATE = '2015-01-01', @ToDate DATE = '2015-06-30';
--DROP Temp TABLE if it exists
IF object_id('tempdb..#GLUK') is not null BEGIN DROP TABLE #GLUK END
-- Select the required GL Transactions by Date Range
/* Create #GLUK */
CREATE TABLE #GLUK (Account NVARCHAR (20),[Year] INT, [Per] VARCHAR(3),[Posting Date] DATE, [Product Series] NVARCHAR (20),[Country] NVARCHAR (20),[Amount] DECIMAL(10,2))
INSERT INTO #GLUK (Account,[Year],[Per],[Posting Date],[Product Series],[Country],[Amount])
SELECT [G_L Account No_],YEAR([Posting Date]) AS [Year],dbo.IDEALfnPER ([Posting Date]) AS [Per],[Posting Date],[Global Dimension 2 Code] AS [Product Series],[Global Dimension 1 Code] AS [Country],[Amount]
FROM [IDEAL_LIVE].[dbo].[UK$G_L Entry] AS GLE_UK
WHERE [Posting Date] >= @FromDate AND [Posting Date] <= @ToDate
--Exclude Inter Company.
AND [G_L Account No_] NOT IN ('52115','51115')
--REVENUE BY PRODUCT SERIES
--DROP Temp TABLE if it exists
IF object_id('tempdb..#PS_REV_UK') is not null BEGIN DROP TABLE #PS_REV_UK END
/* Create #PS_REV_UK */
CREATE TABLE #PS_REV_UK ([Product Series] NVARCHAR (20),[R01] DECIMAL (10,2),[R02] DECIMAL (10,2),[R03] DECIMAL (10,2),[R04] DECIMAL (10,2),[R05] DECIMAL (10,2),[R06] DECIMAL (10,2),
[R07] DECIMAL (10,2),[R08] DECIMAL (10,2),[R09] DECIMAL (10,2),[R10] DECIMAL (10,2),[R11] DECIMAL (10,2),[R12] DECIMAL (10,2))
INSERT INTO #PS_REV_UK ([Product Series],[R01],[R02],[R03],[R04],[R05],[R06],[R07],[R08],[R09],[R10],[R11],[R12])
SELECT [Product Series],
Sum([01])*-1 AS 'R01',Sum([02])*-1 AS 'R02',Sum([03])*-1 AS 'R03',Sum([04])*-1 AS 'R04',
Sum([05])*-1 AS 'R05',Sum([06])*-1 AS 'R06',Sum([07])*-1 AS 'R07',Sum([08])*-1 AS 'R08',
Sum([09])*-1 AS 'R09',Sum([10])*-1 AS 'R10',Sum([11])*-1 AS 'R11',Sum([12])*-1 AS 'R12'
From #GLUK
PIVOT (SUM(Amount) FOR Per in ([01],[02],[03],[04],[05],[06],[07],[08],[09],[10],[11],[12])) AS PTable
WHERE Account LIKE '511%'
GROUP BY [Product Series];
--COST BY PRODUCT SERIES
--DROP Temp TABLE if it exists
IF object_id('tempdb..#PS_COST_UK') is not null BEGIN DROP TABLE #PS_COST_UK END
/* Create #PS_COST_UK */
CREATE TABLE #PS_COST_UK ([Product Series] NVARCHAR (20),[C01] DECIMAL (10,2),[C02] DECIMAL (10,2),[C03] DECIMAL (10,2),[C04] DECIMAL (10,2),[C05] DECIMAL (10,2),[C06] DECIMAL (10,2),
[C07] DECIMAL (10,2),[C08] DECIMAL (10,2),[C09] DECIMAL (10,2),[C10] DECIMAL (10,2),[C11] DECIMAL (10,2),[C12] DECIMAL (10,2))
INSERT INTO #PS_COST_UK ([Product Series],[C01],[C02],[C03],[C04],[C05],[C06],[C07],[C08],[C09],[C10],[C11],[C12])
SELECT [Product Series],
Sum([01]) AS 'C01',Sum([02]) AS 'C02',Sum([03]) AS 'C03',Sum([04]) AS 'C04',
Sum([05]) AS 'C05',Sum([06]) AS 'C06',Sum([07]) AS 'C07',Sum([08]) AS 'C08',
Sum([09]) AS 'C09',Sum([10]) AS 'C10',Sum([11]) AS 'C11',Sum([12]) AS 'C12'
From #GLUK
PIVOT (SUM(Amount) FOR Per in ([01],[02],[03],[04],[05],[06],[07],[08],[09],[10],[11],[12])) AS PTable
WHERE Account LIKE '521%'
GROUP BY [Product Series];
/*****PRODUCT SERIES OUTPUT*****/
--Join the two and calculate the margin.
SELECT RC.[Product Series],
-- Remove NULLS using Function. Required to avoid Divide by Zero Errors
[dbo].[IDEALfnNULLS to Zero](Sum([R01])) AS 'R01',
CAST (CASE WHEN Sum([R01])= 0 OR Sum([R01]) IS NULL OR Sum([C01]) = 0 OR Sum([C01]) IS NULL THEN 0 ELSE
((Sum([R01]) - Sum([C01]))/(Sum([R01])))*100 END AS DECIMAL(10,2)) AS 'M01',
[dbo].[IDEALfnNULLS to Zero](Sum([C01])) AS 'C01',
[dbo].[IDEALfnNULLS to Zero](Sum([R02])) AS 'R02',
CAST (CASE WHEN Sum([R02])= 0 OR Sum([R02]) IS NULL OR Sum([C02]) = 0 OR Sum([C02]) IS NULL THEN 0 ELSE
((Sum([R02]) - Sum([C02]))/(Sum([R02])))*100 END AS DECIMAL(10,2)) AS 'M02',
[dbo].[IDEALfnNULLS to Zero](Sum([C02])) AS 'C02',
[dbo].[IDEALfnNULLS to Zero](Sum([R03])) AS 'R03',
CAST (CASE WHEN Sum([R03])= 0 OR Sum([R03]) IS NULL OR Sum([C03]) = 0 OR Sum([C03]) IS NULL THEN 0 ELSE
((Sum([R03]) - Sum([C03]))/(Sum([R03])))*100 END AS DECIMAL(10,2)) AS 'M03',
[dbo].[IDEALfnNULLS to Zero](Sum([C03])) AS 'C03',
[dbo].[IDEALfnNULLS to Zero](Sum([R04])) AS 'R04',
CAST (CASE WHEN Sum([R04])= 0 OR Sum([R04]) IS NULL OR Sum([C04]) = 0 OR Sum([C04]) IS NULL THEN 0 ELSE
((Sum([R04]) - Sum([C04]))/(Sum([R04])))*100 END AS DECIMAL(10,2)) AS 'M04',
[dbo].[IDEALfnNULLS to Zero](Sum([C04])) AS 'C04',
[dbo].[IDEALfnNULLS to Zero](Sum([R05])) AS 'R05',
CAST (CASE WHEN Sum([R05])= 0 OR Sum([R05]) IS NULL OR Sum([C05]) = 0 OR Sum([C05]) IS NULL THEN 0 ELSE
((Sum([R05]) - Sum([C05]))/(Sum([R05])))*100 END AS DECIMAL(10,2)) AS 'M05',
[dbo].[IDEALfnNULLS to Zero](Sum([C05])) AS 'C05',
[dbo].[IDEALfnNULLS to Zero](Sum([R06])) AS 'R06',
CAST (CASE WHEN Sum([R06])= 0 OR Sum([R06]) IS NULL OR Sum([C06]) = 0 OR Sum([C06]) IS NULL THEN 0 ELSE
((Sum([R06]) - Sum([C06]))/(Sum([R06])))*100 END AS DECIMAL(10,2)) AS 'M06',
[dbo].[IDEALfnNULLS to Zero](Sum([C06])) AS 'C06',
[dbo].[IDEALfnNULLS to Zero](Sum([R07])) AS 'R07',
CAST (CASE WHEN Sum([R07])= 0 OR Sum([R07]) IS NULL OR Sum([C07]) = 0 OR Sum([C07]) IS NULL THEN 0 ELSE
((Sum([R07]) - Sum([C07]))/(Sum([R07])))*100 END AS DECIMAL(10,2)) AS 'M07',
[dbo].[IDEALfnNULLS to Zero](Sum([C07])) AS 'C07',
[dbo].[IDEALfnNULLS to Zero](Sum([R08])) AS 'R08',
CAST (CASE WHEN Sum([R08])= 0 OR Sum([R08]) IS NULL OR Sum([C08]) = 0 OR Sum([C08]) IS NULL THEN 0 ELSE
((Sum([R08]) - Sum([C08]))/(Sum([R08])))*100 END AS DECIMAL(10,2)) AS 'M08',
[dbo].[IDEALfnNULLS to Zero](Sum([C08])) AS 'C08',
[dbo].[IDEALfnNULLS to Zero](Sum([R09])) AS 'R09',
CAST (CASE WHEN Sum([R09])= 0 OR Sum([R09]) IS NULL OR Sum([C09]) = 0 OR Sum([C09]) IS NULL THEN 0 ELSE
((Sum([R09]) - Sum([C09]))/(Sum([R09])))*100 END AS DECIMAL(10,2)) AS 'M09',
[dbo].[IDEALfnNULLS to Zero](Sum([C09])) AS 'C09',
[dbo].[IDEALfnNULLS to Zero](Sum([R10])) AS 'R10',
CAST (CASE WHEN Sum([R10])= 0 OR Sum([R10]) IS NULL OR Sum([C10]) = 0 OR Sum([C10]) IS NULL THEN 0 ELSE
((Sum([R10]) - Sum([C10]))/(Sum([R10])))*100 END AS DECIMAL(10,2)) AS 'M10',
[dbo].[IDEALfnNULLS to Zero](Sum([C10])) AS 'C10',
[dbo].[IDEALfnNULLS to Zero](Sum([R11])) AS 'R11',
CAST (CASE WHEN Sum([R11])= 0 OR Sum([R11]) IS NULL OR Sum([C11]) = 0 OR Sum([C11]) IS NULL THEN 0 ELSE
((Sum([R11]) - Sum([C11]))/(Sum([R11])))*100 END AS DECIMAL(10,2)) AS 'M11',
[dbo].[IDEALfnNULLS to Zero](Sum([C11])) AS 'C11',
[dbo].[IDEALfnNULLS to Zero](Sum([R12])) AS 'R12',
CAST (CASE WHEN Sum([R12])= 0 OR Sum([R12]) IS NULL OR Sum([C12]) = 0 OR Sum([C12]) IS NULL THEN 0 ELSE
((Sum([R12]) - Sum([C12]))/(Sum([R12])))*100 END AS DECIMAL(10,2)) AS 'M12',
[dbo].[IDEALfnNULLS to Zero](Sum([C12])) AS 'C12',
-- TOTALS
[dbo].[IDEALfnNULLS to Zero](Sum([R01]))+[dbo].[IDEALfnNULLS to Zero](Sum([R02]))+[dbo].[IDEALfnNULLS to Zero](Sum([R03]))+[dbo].[IDEALfnNULLS to Zero](Sum([R04]))+[dbo].[IDEALfnNULLS to Zero](Sum([R05]))+[dbo].[IDEALfnNULLS to Zero](Sum([R06]))+
[dbo].[IDEALfnNULLS to Zero](Sum([R07]))+[dbo].[IDEALfnNULLS to Zero](Sum([R08]))+[dbo].[IDEALfnNULLS to Zero](Sum([R09]))+[dbo].[IDEALfnNULLS to Zero](Sum([R10]))+[dbo].[IDEALfnNULLS to Zero](Sum([R11]))+[dbo].[IDEALfnNULLS to Zero](Sum([R12])) AS 'RTotal',
[dbo].[IDEALfnNULLS to Zero](Sum([C01]))+[dbo].[IDEALfnNULLS to Zero](Sum([C02]))+[dbo].[IDEALfnNULLS to Zero](Sum([C03]))+[dbo].[IDEALfnNULLS to Zero](Sum([C04]))+[dbo].[IDEALfnNULLS to Zero](Sum([C05]))+[dbo].[IDEALfnNULLS to Zero](Sum([C06]))+
[dbo].[IDEALfnNULLS to Zero](Sum([C07]))+[dbo].[IDEALfnNULLS to Zero](Sum([C08]))+[dbo].[IDEALfnNULLS to Zero](Sum([C09]))+[dbo].[IDEALfnNULLS to Zero](Sum([C10]))+[dbo].[IDEALfnNULLS to Zero](Sum([C11]))+[dbo].[IDEALfnNULLS to Zero](Sum([C12])) AS 'CTotal'
-- ***** NEED THE TOTAL MARGIN CALC HERE *****
FROM #PS_REV_UK AS RC
LEFT OUTER JOIN #PS_COST_UK AS CC ON CC.[Product Series] = RC.[Product Series]
GROUP BY RC.[Product Series]
Data---Expected-Result.xlsx
Thanks. That will take a little while to work through.
In the meantime please note that the safest of all date literals in SQL Server is YYYYMMDD
e.g. instead of
DECLARE @FromDate DATE = '2015-01-01', @ToDate DATE = '2015-06-30';
use
DECLARE @FromDate DATE = '20150101', @ToDate DATE = '20150701';
& Best practice for date ranges is NOT to use BETWEEN and whilst you haven't literally used this, you are currently using the equivalent.
see: "Beware of Between"
so, instead of this:
WHERE [Posting Date] >= @FromDate AND [Posting Date] <= @ToDate
use:
WHERE [Posting Date] >= @FromDate AND [Posting Date] < @ToDate
The difference is subtle, but it allows the raw data to be in any datetime precision and your query will remain accurate.
In the meantime please note that the safest of all date literals in SQL Server is YYYYMMDD
e.g. instead of
DECLARE @FromDate DATE = '2015-01-01', @ToDate DATE = '2015-06-30';
use
DECLARE @FromDate DATE = '20150101', @ToDate DATE = '20150701';
& Best practice for date ranges is NOT to use BETWEEN and whilst you haven't literally used this, you are currently using the equivalent.
see: "Beware of Between"
so, instead of this:
WHERE [Posting Date] >= @FromDate AND [Posting Date] <= @ToDate
use:
WHERE [Posting Date] >= @FromDate AND [Posting Date] < @ToDate
The difference is subtle, but it allows the raw data to be in any datetime precision and your query will remain accurate.
The approach of using 2 joined temp tables leads to many potential NULLs
I would suggest making it far easier by doing this, into a single temp table if you prefer, or as a "derived table" (which is probably how I would do it):
Note, the row totals are already included.
This approach should result in zeros if there was no values to sum, and from this you should now be able to produce your margin calculations using simple case expressions, even for the totals.
bottom line:
In the "good old days" before "pivot" existed as a TSQL feature we always used case expressions and group by to arrive at pivoted data. In many way it is still the easier method, please don't assume that the "pivot" feature is necessarily the best approach.
& I have also applied the recommended change regarding the date range filter
see: "Beware of Between"
{+ edit}
I overlooked putting " ELSE 0 " in the case expressions. I suggest you include this please.
I would suggest making it far easier by doing this, into a single temp table if you prefer, or as a "derived table" (which is probably how I would do it):
DECLARE @FromDate date = '20150101'
, @ToDate date = '20150701';
SELECT
[Global Dimension 2 Code] AS [Product Series]
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '01' THEN [Amount] END) AS R01
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '02' THEN [Amount] END) AS R02
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '03' THEN [Amount] END) AS R03
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '04' THEN [Amount] END) AS R04
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '05' THEN [Amount] END) AS R05
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '06' THEN [Amount] END) AS R06
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '07' THEN [Amount] END) AS R07
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '08' THEN [Amount] END) AS R08
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '09' THEN [Amount] END) AS R09
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '10' THEN [Amount] END) AS R10
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '11' THEN [Amount] END) AS R11
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' AND CA.Per = '12' THEN [Amount] END) AS R12
, SUM(CASE WHEN [G_L Account No_] LIKE '511%' THEN [Amount] END) AS Rtotal
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '01' THEN [Amount] END) AS C01
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '02' THEN [Amount] END) AS C02
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '03' THEN [Amount] END) AS C03
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '04' THEN [Amount] END) AS C04
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '05' THEN [Amount] END) AS C05
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '06' THEN [Amount] END) AS C06
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '07' THEN [Amount] END) AS C07
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '08' THEN [Amount] END) AS C08
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '09' THEN [Amount] END) AS C09
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '10' THEN [Amount] END) AS C10
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '11' THEN [Amount] END) AS C11
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' AND CA.Per = '12' THEN [Amount] END) AS C12
, SUM(CASE WHEN [G_L Account No_] LIKE '521%' THEN [Amount] END) AS Ctotal
FROM [IDEAL_LIVE].[dbo].[UK$G_L Entry] AS GLE_UK
CROSS APPLY (
SELECT dbo.IDEALfnPER(GLE_UK.[Posting Date])
) CA ([Per])
WHERE [Posting Date] >= @FromDate
AND [Posting Date] < @ToDate
--Exclude Inter Company.
AND [G_L Account No_] NOT IN ('52115', '51115')
GROUP BY
[Global Dimension 2 Code]
Note, the row totals are already included.
This approach should result in zeros if there was no values to sum, and from this you should now be able to produce your margin calculations using simple case expressions, even for the totals.
bottom line:
In the "good old days" before "pivot" existed as a TSQL feature we always used case expressions and group by to arrive at pivoted data. In many way it is still the easier method, please don't assume that the "pivot" feature is necessarily the best approach.
& I have also applied the recommended change regarding the date range filter
see: "Beware of Between"
{+ edit}
I overlooked putting " ELSE 0 " in the case expressions. I suggest you include this please.
Premium Content
You need an Expert Office subscription to comment.Start Free Trial