Question

i have the following table generated by SQL TABLE A

timeinterval   count(exclusive range)
0-6             2
0-12            5
0-18            10

i want a table like this TABLE B

timeinterval   count(exclusive range)  count(inclusive range)
1-6             2                       2
1-12            5                       3
1-18            10                      5

i have already generated table A and need table B. can i do something in SQL where i can add a query in the code for table A and do something like this (0-12)-(0-6) for 2nd row in table B.

code used for generating table A is

with ranges as 
  (
    select 6 as val, 1 as count_all
 union all
    select 12, 1
    union all
    select 18, 1
    union all
 select 24, 1
    union all
 select 30, 1
    union all  
 select 36, 1
    union all 
 select 42, 1
    union all
 select 48, 1
    union all   
 select 1, 0
  )
select case when ranges.count_all = 0
            then 'more'
            else  convert (varchar(10), ranges.val) 
        end [MetLifeExperienceMonths],
       sum (case when (ranges.count_all = 0 and GoldListHistogram.MetLifeExperienceMonths>=1)
                   or
           (GoldListHistogram.MetLifeExperienceMonths<= ranges.val and  GoldListHistogram.MetLifeExperienceMonths>=1)
                 then 1 end) [count],
count(EmployeeID) as 'Total'
into yy
from GoldListHistogram
cross join ranges
where MetLifeExperienceMonths > 0
group by ranges.val, ranges.count_all

i need to modify the query such that i can subtract first two rows value for “count(exclusive range)” for every row staring from the 2nd row..like for 0-12(time interval) row i need to output a value that is difference of the first two rows..like row(i)=count(i)-count(i-1).

first column gives the time interval in 5 years (in months) second column calculates no. of employees in the exclusive range like (0-6,0-12,0-18)..6 ,12,18 being no. of months third column calculates no. of employees in the exclusive range like (0-6,6-12,12-18)

Answer 1

Could you not just add a start value to ranges? Something like:

with ranges as 
  (
    select 6 as val, 0 as start, 1 as count_all
 union all
    select 12, 7, 1
    union all
    select 18, 13, 1
    union all
 select 24, 19, 1
    union all
 select 30, 25, 1
    union all  
 select 36, 31, 1
    union all 
 select 42, 37, 1
    union all
 select 48, 43, 1
    union all   
 select 1, 49, 0
  )
select case when ranges.count_all = 0
            then 'more'
            else  convert (varchar(10), ranges.val) 
        end [MetLifeExperienceMonths],
       sum (case when (ranges.count_all = 0 and GoldListHistogram.MetLifeExperienceMonths>=1)
                   or
           (GoldListHistogram.MetLifeExperienceMonths=1)
                 then 1 else 0 end) [count inclusive],
       sum (case when (ranges.count_all = 0 and GoldListHistogram.MetLifeExperienceMonths>=1)
                   or
           (GoldListHistogram.MetLifeExperienceMonths=ranges.start)
                 then 1 else 0 end) [count exclusive],
count(EmployeeID) as 'Total'
into yy
from GoldListHistogram
cross join ranges
where MetLifeExperienceMonths > 0
group by ranges.val, ranges.count_all;