sql - dynamically populate columns in a result set based on another column - sql

Question

We want to determine the amount of assets that customers have insured or uninsured.
Basically the insured amount is £85K, and we aportion that accross several account types held by the custimer in order of the accounts liquidity, so for example, a current account is more liquid than a 95 day notice account therefore the current account would receive the money first...
We have the below code that generates the priorities based on the account type and of course we could now loop through the table and determine the insured amount for each account type for each customer, and anything non-insured would become the uninsured amount for that customer.
But is there a way to do this in the main select statement?
declare #Priorities TABLE
(
[refID] [int] IDENTITY(1,1) NOT NULL,
acno nvarchar(10),
Suffix nvarchar(6) null,
Unit nvarchar(5) null,
CT nvarchar(2) null,
LOB nvarchar(2) null,
Participants int null,
VAL float null,
xPriority int null,
insuredAmt float null,
uninsuredAmt float null
)
declare #Count int=0
declare #I int=0
insert into #Priorities
select acno, suffix, unit, CT, LOB, Participants, Val, xPriority =
case
when LOB BETWEEN 'GA' AND 'GN' then 1 -- Current accounts
when LOB IN ( 'HY' , 'H1') then 2 -- Call
when CHARINDEX(LOB, 'HA; HC; HM; HQ; IA; IB; IC; IE;IG; IU;')>0 then 3 -- Instant Access
when LOB='HO' then 4 -- 14 Day notice
when CHARINDEX(LOB, 'DN; HG; HK; HL; HP; HS; HU; IV; IW;')>0 then 5 -- 35-Day notice
when LOB IN ('DN', 'HV') then 6 -- 95-Day notice
when (LOB BETWEEN 'IS' AND 'IT') or (LOB between 'JA' and 'JZ' ) then 7 -- Residual Deals
end,
insuredAmt=null,
uninsuredAmt=null
from actdata
where datadate = '20131212' /*#processdate*/
and CHARINDEX(CT, 'AA;AC;AE;AG;BA;BC;BE;BG;CA;DE;DG;ZA;')=0
and LOB BETWEEN 'GA' AND 'KZ' And val>0
and participants is not null
order by acno
what I'd like to do is calculate the insured amount for each record based on the value of the column xPriority.
Also, we multiply the total insured amount by the number of participants in the account, so if there are 5 participants in the account, then the total insured amount for them all would be 5*85000 = £425000.
the alternative would be a loop or a cursor.
this would be run on a daily basis when the data is brought into the reporting database from the iSeries system by another process.
thanks
Philip

Answer 1

I believe the answer is 'No', but that's based on a particular interpretation of:
"Also, we multiply the total insured amount by the number of participants in the account, so if there are 5 participants in the account, then the total insured amount for them all would be 5*85000 = £425000."
Without multiple participants in an account, I think it would be possible to order the SELECT statement by the customer ID and the CASE statement for xPriority, keep a decreasing running total (down form 85K and resetting for each customer) and calculate insuredAmt and uninsuredAmt accordingly. This would tell you how each individual's insured amount was apportioned across their accounts.
However, with multiple participants you have to resolve the following problem:
Mary has a current account of £40k and a 30 day notice account with £20k.
Bob has a current account of £50k and a 30 day notice account with £20k.
Mary and Bob also have a joint, 14 day notice account of £60k.
The two current accounts are fully insured, as is the 14 day notice account (Mary has £45k and Bob £35k allowance left). The problem is that the coverage of the two 30 day accounts depends on which individual's insurance is apportioned to the joint account (all of Mary's plus a bit of Bob's; all of Bob's plus a bit of Mary's; £30k of each; 9:7 ratio from Mary:Bob?). I don't think this problem is even resolvable in logical terms without reference to the full terms and conditions of the accounts! When it comes to programming it, I'd advise extreme caution. Start by writing inefficient code you can test thoroughly on such cases, then use that as a benchmark for checking code with enhanced efficiency.
Good luck!