Add and Integrate 3 fields to a Pivot query

Question

I kept asking questions about this query even after I had marked the thread as answered, and may not have been fair to those who assisted.
Given the following code, how do I add 2 calculated fields:
1. Sales - SalesLy as Difference
2. ((Sales - SalesLY) / SalesLY) * 100 as SalesPercent
This percent should be 2 digits after decimal point and also check for divide by zero.
select oracle_key  
-- "pivot" results
,      max(businessdate)   businessdate
,      max(sales)          sales
,      max(sales_ly)       sales_ly
,      max(guests)         guests
,      max(guests_ly)      guests_ly
,      max(dollarvariance) dollarvariance
,      max(wtd_sales)      wtd_sales
,      max(wtd_sales_ly)   wtd_sales_ly
,      max(wtd_guests)     wtd_guests
,      max(wtd_guests_ly)  wtd_guests_ly
from ( with -- just touching both tables once instead of four times here (subquery factoring)
           totals as ( select d.busi_date
                       ,      d.oracle_key
                       ,      s.storeid
                       ,      d.field1
                       ,      d.field2
                       ,      d.field3
                       -- ,      d.field4 (Cars) doesnt't seem to be part of your final result set?
                       from   mydailytotals d
                       ,      mystores s
                       where  d.oracle_key = s.storeid(+) --<< forgive my 'old school' outer-join ;)
       select oracle_key  
--       ,      storeid
--       ,      row_number() over (partition by oracle_key order by busi_date desc) rn      
       -- note: assuming dates are/will become parameters/bind variables in real code...
       -- businesdate
       ,      case
                when busi_date = to_date('6/4/2009', 'mm/dd/yyyy')
                then busi_date
              end as businessdate
       -- sales             
       ,      case
                when busi_date = to_date('6/4/2009', 'mm/dd/yyyy')
                then field1
              end as sales
       -- saleslastyear             
       ,      case
                when busi_date = to_date('6/5/2008' , 'mm/dd/yyyy')
                then field1
              end as sales_ly
       -- guestcount             
       ,      case
                when busi_date = to_date('6/4/2009', 'mm/dd/yyyy')
                then field2
              end as guests
       -- guestcountlastyear             
       ,      case      
                when busi_date = to_date('6/5/2008', 'mm/dd/yyyy')
                then field2              
              end as guests_ly
       -- dollarvariance
       ,      case
                when busi_date = to_date('6/4/2009', 'mm/dd/yyyy')
                then field3
              end as dollarvariance
       -- weektodatesales             
       ,      sum( case
                     when busi_date between to_date('6/1/2009', 'mm/dd/yyyy') and to_date('6/4/2009', 'mm/dd/yyyy')
                     then field1 else 0
                   end
              as wtd_sales
       -- weektodatesaleslastyear             
       ,      sum( case
                     when busi_date between to_date('6/2/2008', 'mm/dd/yyyy') and to_date('6/5/2008', 'mm/dd/yyyy')
                     then field1 else 0
                   end
              ) as wtd_sales_ly
       -- weektodateguestcount             
       ,      sum( case
                     when busi_date between to_date('6/1/2009', 'mm/dd/yyyy') and to_date('6/4/2009', 'mm/dd/yyyy')
                     then field2 else 0  
                   end
              ) as wtd_guests
       -- weektodateguestcountlastyear             
       ,      sum( case
                     when busi_date between to_date('6/2/2008', 'mm/dd/yyyy') and to_date('6/5/2008', 'mm/dd/yyyy')
                    then field2 else 0
                   end
              ) as wtd_guests_ly
       from totals
       group by oracle_key
       ,        busi_date      
       ,        field1
       ,        field2
       ,        field3      
group by  oracle_key
order  by oracle_keySee this post for background:
Complex Inner Join and subquery on outer where clause on inner join?
Another question regarding the same. The following query successfully eliminates duplicates in the DollarVariance field.
How do I integrate into previous query.
Select StoreId,
           DollarVariance
    from   (select store.storeid,
                    case
                          when inv.fk_str_main_id is null then 'None'
                          when inv.is_posted = 0 then 'NP'
                          else chr(39)||inv.total_dol_var||chr(39)
                     end as DollarVariance ,
               row_number() over (partition by store.storeid order by inv.is_posted desc) rn 
               from  valenti_wendys.inv_main inv
               right outer join valenti_wendys.str_main_mapped_to_storedb store
                              on store.storeid = inv.fk_str_main_id
                              and FK_LK_CNT_FREQ_ID = 1
                              and busi_date = PrevDay)
      where rn = 1;If more information needed, refer to this post:
URLhttp://forums.oracle.com/forums/thread.jspa?threadID=912671&tstart=60 [URL]
Thanks in advance

Answer 1

Yeah, I know, I'm a bit later than planned, about to log off, but I know ;)
How about:
select oracle_key  
-- "pivot" results
,      max(businessdate)   businessdate
,      max(sales)          sales
,      max(sales_ly)       sales_ly
,      max(sales) - max(sales_ly) difference
,      round(((max(sales) - max(sales_ly))/max(sales_ly)*100), 2) salespercent
,      max(guests)         guests
,      max(guests_ly)      guests_ly
,      max(dollarvariance) dollarvariance
,      max(wtd_sales)      wtd_sales
,      max(wtd_sales_ly)   wtd_sales_ly
,      max(wtd_guests)     wtd_guests
,      max(wtd_guests_ly)  wtd_guests_ly
from ( with -- just touching both tables once instead of four times here (subquery factoring)
           totals as ( select d.busi_date
                       ,      d.oracle_key
                       ,      s.storeid
                       ,      d.field1
                       ,      d.field2
                       ,      d.field3
                       -- ,      d.field4 (Cars) doesnt't seem to be part of your final result set?
                       from   mydailytotals d
                       ,      mystores s
                       where  d.oracle_key = s.storeid(+) --<< forgive my 'old school' outer-join ;)
       select oracle_key  
--       ,      storeid
--       ,      row_number() over (partition by oracle_key order by busi_date desc) rn      
       -- note: assuming dates are/will become parameters/bind variables in real code...
       -- businesdate
       ,      case
                when busi_date = to_date('6/4/2009', 'mm/dd/yyyy')
                then busi_date
              end as businessdate
       -- sales             
       ,      case
                when busi_date = to_date('6/4/2009', 'mm/dd/yyyy')
                then field1
              end as sales
       -- saleslastyear             
       ,      case
                when busi_date = to_date('6/5/2008' , 'mm/dd/yyyy')
                then field1
              end as sales_ly
       -- guestcount             
       ,      case
                when busi_date = to_date('6/4/2009', 'mm/dd/yyyy')
                then field2
              end as guests
       -- guestcountlastyear             
       ,      case      
                when busi_date = to_date('6/5/2008', 'mm/dd/yyyy')
                then field2              
              end as guests_ly
       -- dollarvariance
       ,      case
                when busi_date = to_date('6/4/2009', 'mm/dd/yyyy')
                then field3
              end as dollarvariance
       -- weektodatesales             
       ,      sum( case
                     when busi_date between to_date('6/1/2009', 'mm/dd/yyyy') and to_date('6/4/2009', 'mm/dd/yyyy')
                     then field1 else 0
                   end
              as wtd_sales
       -- weektodatesaleslastyear             
       ,      sum( case
                     when busi_date between to_date('6/2/2008', 'mm/dd/yyyy') and to_date('6/5/2008', 'mm/dd/yyyy')
                     then field1 else 0
                   end
              ) as wtd_sales_ly
       -- weektodateguestcount             
       ,      sum( case
                     when busi_date between to_date('6/1/2009', 'mm/dd/yyyy') and to_date('6/4/2009', 'mm/dd/yyyy')
                     then field2 else 0  
                   end
              ) as wtd_guests
       -- weektodateguestcountlastyear             
       ,      sum( case
                     when busi_date between to_date('6/2/2008', 'mm/dd/yyyy') and to_date('6/5/2008', 'mm/dd/yyyy')
                    then field2 else 0
                   end
              ) as wtd_guests_ly
       from totals
       group by oracle_key
       ,        busi_date      
       ,        field1
       ,        field2
       ,        field3      
group by  oracle_key
order  by oracle_key
notez bien
+(still) using the data from your original post here+
Is it not possible, that since you have the fields available in the inner query, you could avoid the need for another? Querying a result set containing all the data you need is less resource intensive than requerying the same data/tables over and over.
A common way to create reports is to base it on a (smart/dedicated) view, the view would deliver the required output already JAPYR.
I have a program that produces report already, but in the business logic.Please elaborate on that, so we can understand better what's goin' on.
Another question regarding the same. The following query successfully eliminates duplicates in the DollarVariance field.
How do I integrate into previous query. I'll get back on that one tomorrow
Edited by: hoek on Jun 21, 2009 11:06 PM