Join tables on multiple conditions where rows may not have all IDs - sql

Question

I need to build a query for a report that will return both budget and transaction data for all account types. The problem is that there are some account types present in the budget table that are not in the transaction table and vice versa. Here are the queries with the relevant information that needs to be joined from each table. First, the budget:
Select bp.ProjectID, bp.AccountCategoryID, bp. FiscalYearPeriod, bp.Amount AS BudgetAmount
from BudgetsProject bp
where bp.ProjectId = 1063 and bp.FiscalYearPeriod = 201510
The relevant data from the transaction table is found using:
SELECT f.ProjectKey, f.AccountCategoryKey, f.PostFiscalPeriodKey, sum(f.IncomeAmount) as IncomeAmount, sum(f.ExpenseAmount) as ExpenseAmount, sum(f.TransactionCount) as TransactionCount
FROM ActualCategoryPivotAccountTypeSubAccountFact f
Where f.ProjectKey = 1063 and f.PostFiscalPeriodKey = 201510
Group by f.ProjectKey, f.AccountCategoryKey, f.PostFiscalPeriodKey
The problem is that when I attempt to FULL JOIN the tables not every AccountCategoryKey value is in every row for both tables so I end up missing rows of data. I need to have every possible Accountcategory show in the result set and have NULLs where corresponding transaction or budget isn't available.
I know that this answer exists: What kind of join do I need to use to create one table from many?
but the tables must be joined on matching Project, AccountCategory, and FiscalYear fields and I'm not sure how to apply the solution presented there to a case where I need to be joining on multiple conditions.
EDIT: If this is helpful I've added logic to recreate tables here:
Create Table BudgetsReport (
AccountCategoryID int NOT NULL,
ProjectID int NOT NULL,
FiscalYearPeriod char(6) NOT NULL,
Amount numeric(10,2)
);
Create Table ActualCategoryPivotAccountTypeSubAccountFact (
PostFiscalPeriodKey int,
ProjectKey int,
AccountCategoryKey int,
IncomeAmount money,
ExpenseAmount money,
TransactionCount int
);
INSERT INTO BudgetsReport VALUES (1063, 5, 201510, 1626.00), (1063, 15, 201510, 8.00), (1063, 26, 201510, 1757.00), (1063, 36, 201510, 0.00), (1063, 38, 201510, 6508.00), (1063, 41, 201510, 115000.00), (1063, 42, 201510, 667.00), (1063, 43, 201510, 167.00), (1063, 51, 201510, 7289.00), (1063, 54, 201510, 21.00), (1063, 81, 201510, 138164.00), (1063, 87, 201510, 83.00), (1063, 88, 201510, 54.00), (1063, 90, 201510, 833.00);
INSERT INTO ActualCategoryPivotAccountTypeSubAccountFact VALUES (1063, 2, 201510, 0.00, 0.00, 214), (1063, 5, 201510, 0.00, 1004.42, 1), (1063, 15, 201510, 0.00, 3.92, 1), (1063, 26, 201510, 0.00, 1556.44, 10), (1063, 34, 201510, 0.00, 26.37, 1), (1063, 36, 201510, 0.00, 0.00, 19), (1063, 38, 201510, 0.00, 5764.65, 10), (1063, 41, 201510, 0.00, 131857.10, 29), (1063, 51, 201510, 0.00, 6456.27, 10), (1063, 54, 201510, 0.00, .44, 1), (1063, 87, 201510, 0.00, 28.30, 1), (1063, 90, 201510, 0.00, 545.96, 120), (1063, 93, 201510, 149945.01, 0.00, 213);

Answer 1

Having those predicates in your where clause (WHERE f.PostFiscalPeriodKey = 201510 and f.ProjectKey = 1063) are what's killing you, I think. You could move them to the on clause, but with a full join I don't know what that might end up with. Maybe try using a derived table with your constraints, and then join to that.
select
...
FROM
(
select *
from
ActualCategoryPivotAccountTypeSubAccountFact
where
PostFiscalPeriodKey = 201510
and ProjectKey = 1063) F
full join BudgetsProject bp
on f.PostFiscalPeriodKey = bp.FiscalYearPeriod
and bp.ProjectId = f.ProjectKey
and bp.AccountCategoryId = f.AccountCategoryKey
You could also try building a driver table with all the possible keys, and then outer joining to it. Not sure how this would perform though.

Answer 2

I suggest using a CTE and storing the possible account types and periods. Then left join to the actuals and budgets.
Another option would be to create a temp table with those values first then perform the select operation.