Join two tables with different field value - sql-server

Question

Would it be possible to join two tables with different field values? I am using MS SQL Server 2008.
Table A (original image, thanks to #EdwardRusu for translating images to text):
+ ----- + --------- + -------- + ----------- + ------------------ + ---------------- + ----------- +
| RecID | Member ID | LoanType | LoanSubType | Application Number | Application Date | Loan Amount |
+ ----- + --------- + -------- + ----------- + ------------------ + ---------------- + ----------- +
| 3 | 00005 | Regular | | 201604002 | 2016-02-28 | 39864.00 |
| 185 | 00005 | Special | Special ... | 201604183 | 2016-10-31 | 10000.00 |
| 318 | 00005 | Regular | | 201605063 | 2016-05-18 | 39864.00 |
| 427 | 00005 | Regular | | 201608021 | 2016-08-18 | 39872.00 |
| 486 | 00005 | Special | Special ... | 201609044 | 2016-09-07 | 10000.00 |
| 589 | 00005 | Regular | | 201611008 | 2016-11-04 | 39872.00 |
| 689 | 00005 | Regular | | 201702004 | 2017-02-02 | 39872.00 |
+ ----- + --------- + -------- + ----------- + ------------------ + ---------------- + ----------- +
Table B (original image):
+ --------------- + --------- + ------ + ----- + ---------- + -------- + -------- + -------- +
| ProjectAcctCode | Member ID | TMonth | TYear | TLastDate | TDebit | TCredit | TBalance |
+ --------------- + --------- + ------ + ----- + ---------- + -------- + -------- + -------- +
| 105350500 | 00005 | 1 | 2017 | 2017-01-31 | 0.00 | 2952.00 | -2952.00 |
| 105350500 | 00005 | 5 | 2016 | 2016-05-31 | 73084.00 | 33220.00 | 39864.00 |
| 105350500 | 00005 | 6 | 2016 | 2016-06-30 | 0.00 | 2951.42 | -2952.42 |
| 105350500 | 00005 | 7 | 2016 | 2016-07-31 | 0.00 | 3014.14 | -3014.14 |
| 105350500 | 00005 | 8 | 2016 | 2016-08-31 | 39872.00 | 33905.26 | 5973.55 |
| 105350500 | 00005 | 9 | 2016 | 2016-09-30 | 0.00 | 2952.00 | -2952.00 |
| 105350500 | 00005 | 10 | 2016 | 2016-10-31 | 0.00 | 3014.73 | -3014.73 |
| 105350500 | 00005 | 11 | 2016 | 2016-11-30 | 39872.00 | 33905.26 | 5966.74 |
| 105351000 | 00005 | 1 | 2017 | 2017-01-31 | 0.00 | 975.03 | -975.03 |
| 105351000 | 00005 | 5 | 2016 | 2016-05-31 | 5000.00 | 1000.00 | 4000.00 |
| 105351000 | 00005 | 6 | 2016 | 2016-06-30 | 0.00 | 1000.00 | -1000.00 |
| 105351000 | 00005 | 7 | 2016 | 2016-07-31 | 0.00 | 1000.00 | -1000.00 |
| 105351000 | 00005 | 8 | 2016 | 2016-08-31 | 0.00 | 1000.00 | -1000.00 |
| 105351000 | 00005 | 9 | 2016 | 2016-09-30 | 10000.00 | 1000.00 | 9000.00 |
| 105351000 | 00005 | 10 | 2016 | 2016-10-31 | 0.00 | 955.82 | -955.82 |
| 105351000 | 00005 | 11 | 2016 | 2016-11-30 | 0.00 | 965.38 | -965.38 |
+ --------------- + --------- + ------ + ----- + ---------- + -------- + -------- + -------- +
I want to get the total TBalance from TableB grouped by MemberID, LoanType, ApplicationNo, ApplicationDate. These two tables should be joined based on LoanType and ProjAcctCode AND ApplicationDate and TLastDate. LoanType and ProjAcctCode have different values. But, "REGULAR" Loantype is equal to "105350500" ProjAcctCode and "SPECIAL LoanType" is equivalent to "105351000" ProjAcctCode.
ApplicationDate should be less than or equal to TLastDate.
So if I will generate records for "REGULAR" loan types , I should have something like this (original image):
+ -------- + -------- + ------------- + --------------- + ---------- + ------------ + --------- + --------- +
| MemberID | LoanType | ApplicationNO | ApplicationDate | LoanAmount | ProjAcctCode | TLastDate | Balance |
+ -------- + -------- + ------------- + --------------- + ---------- + ------------ + --------- + --------- +
| 000005 | Regular | 201608021 | 8/18/2016 | 39,872.00 | 105350500 | 8/31/2016 | 39,871.44 |
+ -------- + -------- + ------------- + --------------- + ---------- + ------------ + --------- + --------- +
But with my query,
SELECT a.MemberID,
(SELECT TOP (1) ApplicationNo
FROM TABLE A
WHERE (MemberID = a.MemberID) AND (ApplicationDate <= b.TLastDate)
ORDER BY ApplicationNo DESC) AS ApplicationNo,
(SELECT TOP (1) LoanAmount
FROM TABLE A AS SAL_APPLICATION_HEADER_1
WHERE (MemberID = a.MemberID) AND (ApplicationDate <= b.TLastDate)
ORDER BY ApplicationNo DESC) AS LoanAmount,
(SELECT TOP (1) ApplicationDate
FROM TABLE A AS SAL_APPLICATION_HEADER_2
WHERE (MemberID = a.MemberID)
AND (ApplicationDate <= b.TLastDate)
ORDER BY ApplicationNo DESC) AS ApplicationDate,
vwSAL_Balance_SL_1.ProjAcctCode,
b.TDebit,
b.TCredit,
b.TBalance AS Balance,
b.TLastDate
FROM TABLE A AS a
INNER JOIN TABLE B AS b ON a.MemberID = b.SLCode
GROUP BY a.MemberID,
b.TDebit,
b.TCredit,
b.TBalance,
b.ProjAcctCode,
b.TLastDate
HAVING (a.MemberID = N'00005') AND (b.TLastDate = '8/31/2016')
I got this result (original image):
+ -------- + ------------- + ---------- + --------------- + ------------ + -------- + -------- + -------- + ---------- +
| MemberID | ApplicationNo | LoanAmount | ApplicationDate | ProjAcctCode | TDebit | TCredit | Balance | TLastDate |
+ -------- + ------------- + ---------- + --------------- + ------------ + -------- + -------- + -------- + ---------- +
| 00005 | 201608021 | 39872.00 | 2016-08-18 | 105351000 | 0.00 | 1000.00 | -1000.00 | 2016-08-31 |
| 00005 | 201608021 | 39872.00 | 2016-08-18 | 105350500 | 39872.00 | 33898.45 | 5973.55 | 2016-08-31 |
+ -------- + ------------- + ---------- + --------------- + ------------ + -------- + -------- + -------- + ---------- +
This might be too long but please help. Thank you.

Answer 1

tl;dr Yes, it is possible. The query block at the end will give you what you want.
Yes, it is possible to join two tables with different field values as long as you know some sort of correlation between the two. I will walk you through the construction of the join statement that gives you what you want. You need three things:
1) Match MemberID and SLCode
2) Match LoanType and ProjAcctCode
3) Match the dates
MemberID and SLCode
Based on your comment above, I'm going to assume that SLCode is the same thing as MemberID in B.
select *
from A
inner join B
on B.MemberID = A.MemberID -- or B.SLCode = A.MemberID
LoanType and ProjAcctCode
In this example, you want to use a case statement in the join, something like
inner join B.ProjAcctCode = case A.LoanType when 'Regular' then 105350500 else 105351000 end
This says that if the LoanType is Regular, then only join records that have ProjAcctCode = 105350500; otherwise join records that have ProjAcctCode 105351000. (Here, LoanType only has two states, so using the else clause to capture Special loans is perfectly fine. If there are more than two states, then you must use additional conditions).
select *
from A
inner join B
on B.MemberID = A.MemberID
and B.ProjAcctCode = case A.LoanType when 'Regular' then 105350500 else 105351000 end
Dates
Getting the date filter right is immediate.
select *
from A
inner join B
on B.MemberID = A.MemberID
and B.ProjAcctCode = case A.LoanType when 'Regular' then 105350500 else 105351000 end
and A.ApplicationDate <= B.TLastDate
Group and Aggregate
Now that you have the tables joined appropriately, you just need to group and aggregate. You can group on anything from A since those fields don't change for the aggregates you want, and you just have to be a little clever in how you select things that came from B. In your question, it looks like you want ProjAcctCode (this doesn't change, so we can group on this too), TLastDate (this does change, so we need to aggregate on some criteria), and TBalance (from which we want the sum). I'm going to assume that you want the latest TLastDate, so your query will be something like
select A.MemberID,
A.LoanType,
A.ApplicationNumber,
A.ApplicationDate,
[anything else you want from A],
B.ProjAcctCode -- aggregate not needed because this doesn't change
max(B.TLastDate) as TLastDate,
sum(B.TBalance) as TBalance
from A
inner join B
on B.MemberID = A.MemberID
and B.ProjAcctCode = case A.LoanType when 'Regular' then 105350500 else 105351000 end
and A.ApplicationDate <= B.TLastDate
group by A.MemberID,
A.LoanType,
A.ApplicationNumber,
A.ApplicationDate,
[anything else you selected from A],
B.ProjAcctCode
This produces the following table
+ -------- + -------- + ----------------- + --------------- + ------------ + ---------- + -------- +
| MemberID | LoanType | ApplicationNumber | ApplicationDate | ProjAcctCode | TLastDate | TBalance |
+ -------- + -------- + ----------------- + --------------- + ------------ + ---------- + -------- +
| 00005 | Regular | 201604002 | 2016-02-08 | 105350500 | 2017-01-31 | 36913.19 |
| 00005 | Special | 201604183 | 2016-10-31 | 105351000 | 2017-01-31 | -2896.23 |
| 00005 | Regular | 201605063 | 2016-05-18 | 105350500 | 2017-01-31 | 36913.19 |
| 00005 | Regular | 201608021 | 2016-08-18 | 105350500 | 2017-01-31 | 3014.75 |
| 00005 | Special | 201609044 | 2016-09-07 | 105351000 | 2017-01-31 | 6103.77 |
| 00005 | Regular | 201611008 | 2016-11-04 | 105350500 | 2017-01-31 | 3014.74 |
+ -------- + -------- + ----------------- + --------------- + ------------ + ---------- + -------- +
Note: The loan with ApplicationNumber 201702004 doesn't appear in the final result because it gets filtered out by date (i.e. its application date is greater than all TLastDate in B).

Answer 2

I just use UNION (per LoanType) in my query to get the desire result.