Need Help to Densify and Rank Data

Question

Hi Pros!
I've been researching, but can still use some help here with a telecom application. Say I have the following records in table cost:
Acct,Code,     Rate
1,     001,      .1
1,     0012,      .1
2,     001,      .2
2,     0012,      .2
2,     00123,      .2
2,     001234,     .2
And I want the following results:
Row, Code, Acct, Rate, Rank,
1, 011, 1, .1, 1
2, 011, 2, .2, 2
3, 0012, 1, .1, 1
4, 0012, 2, .2, 2
5, 00123, 1, .1, 1
6, 00123, 2, .2, 2
7, 001234, 1, .1, 1
8, 001234, 2, .2, 2
I use the following query for the ranking:
SELECT code, acct, rate, drank
FROM (SELECT code, acct_id, DENSE_RANK() OVER(PARTITION BY code ORDER BY rate ASC) drank
FROM costs)
Because the data is not uniform (or dense), I would only get the following results:
Row Code Acct Rate Rank
1, 011, 1, .1, 1
2, 011, 2, .2, 2
3, 0012, 1, .1, 1
4, 0012, 2, .2, 2
5, 00123, 2, .2, 1
6, 001234, 2, .2, 1
So what I'm looking for is to densify the data so that for code 00123 and 001234
rates in account 1 would be duplicated for the higher codes as follows:
Acct Code Rate
1, 001, .1
1, 0012, .1
1, 00123, .1
1, 001234, .1
So if accounts without rates for the more specific codes are replicated out to match accounts with the most specific codes, I would have correct ranking. This is for routing purposes, so if a shorter code exists within a longer code and has a cheaper rate, then I would want to have a match there. Cost table can have 10 m rows or more.
Any help would be greatly appreciated
Thanks all.

Answer 1

You may use partitioned outer join.
Completion of data series by analytical function