## T-SQL Calculation

 Author Message PSB SSCertifiable Group: General Forum Members Points: 6793 Visits: 1653 Hi,I have a table where I need to calculate Trend based on Week Type and Unit .CREATE TABLE #REport(TYPE VARCHAR(20),Unit VARCHAR(20),Forecast INT,Week INT,Trend NUMERIC(9,2))INSERT INTO #REport( TYPE, Unit, Forecast, Week, Trend )SELECT 'Net Revenue','XXX',0,1,0UNIONSELECT 'Net Revenue','XXX',0,2,0UNIONSELECT 'Net Revenue','XXX',0,3,0UNIONSELECT 'Net Revenue','XXX',0,4,0UNIONSELECT 'Net Revenue','XXX',105000,5,0UNIONSELECT 'Invest','XXX',0,1,0UNIONSELECT 'Invest','XXX',0,2,0UNIONSELECT 'Invest','XXX',0,3,0UNIONSELECT 'Invest','XXX',0,4,0UNIONSELECT 'Invest','XXX', 45951 ,5,0SELECT * FROM #REportDROP TABLE #REport--FOR every TYPE AND Unit, the trend should be calculated AS follows :--FOR Week 1 OF TYPE = 'Invest' and Unit = 'XXX'--Trend = 45951/MAX week number = 45951/5 = 9190.2--FOR Week 2 OF TYPE = 'Invest' and Unit = 'XXX'--Trend = 9190.2 + 45951/5 = 18380.4-- Trend of Week 1 + 45951/5--FOR Week 3 OF TYPE = 'Invest' and Unit = 'XXX'--Trend = 18380.4 + 45951/5 = 27570.6-- Trend of Week 2 + 45951/5--FOR Week 4 OF TYPE = 'Invest' and Unit = 'XXX'--Trend = 27570.6 + 45951/5 = 36760.8-- Trend of Week 3 + 45951/5--FOR Week 5 OF TYPE = 'Invest' and Unit = 'XXX'--Trend = 36760.8 + 45951/5 = 45951-- Trend of Week 4 + 45951/5Thanks,PSB J Livingston SQL One Orange Chip Group: General Forum Members Points: 27142 Visits: 41336 just to be sure.....can you please provide the actual results you require based on your sample data......and not the calcs.ta. ________________________________________________________________you can lead a user to data....but you cannot make them think and remember....every day is a school day PSB SSCertifiable Group: General Forum Members Points: 6793 Visits: 1653 Desired result ..SELECT 'Invest' AS TYPE,'XXX' AS Unit,0 AS ForeCast,1 AS Week,9190.2 AS TrendUNIONSELECT 'Invest' AS Type,'XXX' AS Unit,0 AS ForeCast,2 AS Week,18380.4 AS TrendUNIONSELECT 'Invest' AS TYPE,'XXX' AS Unit,0 AS ForeCast,3 AS Week,27570.6 AS TrendUNIONSELECT 'Invest' AS Type,'XXX' AS Unit,0 AS ForeCast,4 AS Week,36760.8 AS TrendUNIONSELECT 'Invest' AS Type,'XXX' AS Unit, 45951 AS ForeCast,5 AS Week,45951 AS Trend ChrisM@Work SSC Guru Group: General Forum Members Points: 96694 Visits: 20696 Otherwise you will get answers like this:`SELECT *, Trend = ([Week]/5.0) * MAX(Forecast) OVER(PARTITION BY [TYPE])FROM #REport`Which exactly meets your requirements (apart from the trivial matter of rounding) but is almost certainly incorrect. “Write the query the simplest way. If through testing it becomes clear that the performance is inadequate, consider alternative query forms.” - Gail ShawFor fast, accurate and documented assistance in answering your questions, please read this article.Understanding and using APPLY, (I) and (II) Paul White Hidden RBAR: Triangular Joins / The "Numbers" or "Tally" Table: What it is and how it replaces a loop Jeff ModenExploring Recursive CTEs by Example Dwain Camps J Livingston SQL One Orange Chip Group: General Forum Members Points: 27142 Visits: 41336 ChrisM@Work (5/8/2014)Otherwise you will get answers like this:`SELECT *, Trend = ([Week]/5.0) * MAX(Forecast) OVER(PARTITION BY [TYPE])FROM #REport`Which exactly meets your requirements (apart from the trivial matter of rounding) but is almost certainly incorrect.Chris ...your mind reading skills are quite extraordinary :-)but have a suspicion that OP has oversimplified........we can wait and see ________________________________________________________________you can lead a user to data....but you cannot make them think and remember....every day is a school day ChrisM@Work SSC Guru Group: General Forum Members Points: 96694 Visits: 20696 J Livingston SQL (5/8/2014)ChrisM@Work (5/8/2014)Otherwise you will get answers like this:`SELECT *, Trend = ([Week]/5.0) * MAX(Forecast) OVER(PARTITION BY [TYPE])FROM #REport`Which exactly meets your requirements (apart from the trivial matter of rounding) but is almost certainly incorrect.Chris ...your mind reading skills are quite extraordinary :-)but have a suspicion that OP has oversimplified........we can wait and seeJ Livingstone - you give me way too much credit for a Wild Guess :-D “Write the query the simplest way. If through testing it becomes clear that the performance is inadequate, consider alternative query forms.” - Gail ShawFor fast, accurate and documented assistance in answering your questions, please read this article.Understanding and using APPLY, (I) and (II) Paul White Hidden RBAR: Triangular Joins / The "Numbers" or "Tally" Table: What it is and how it replaces a loop Jeff ModenExploring Recursive CTEs by Example Dwain Camps PSB SSCertifiable Group: General Forum Members Points: 6793 Visits: 1653 Thanks for the solution! it works perfectly .