Query to group by based on partial count match

Question

Dear experts,
Here is a challenge question.
I have a table a on Oracle 11i – this is how it is defined.
desc a
Name Null Type
STORE_NUMBER NOT NULL CHAR(9)
ASSORTMENT_NUMBER NOT NULL CHAR(9)
ITEM_NUMBER NOT NULL CHAR(9)
DESCRIPTION CHAR(1000)
UPC_CODE NOT NULL CHAR(15)
STOCK_FLAG CHAR(1)
AVG_RTL_PRICE CHAR(16)
SALES_MOVEMENT CHAR(16)
PROD_STYLE CHAR(50)
COLOR CHAR(50)
ATT_1 CHAR(50)
ATT_2 CHAR(50)
ATT_3 CHAR(50)
ATT_4 CHAR(50)
ATT_5 CHAR(50)
ATT_6 CHAR(50)
ATT_7 CHAR(50)
ATT_8 CHAR(50)
ATT_9 CHAR(50)
ATT_10 CHAR(50)
ATT_11 CHAR(50)
ATT_12 CHAR(50)
ATT_13 CHAR(50)
ATT_14 CHAR(50)
ATT_15 CHAR(50)
25 rows selected
Primary Key as Store_number, Assortment_number, item_number. Therefore, for a particular assortment number, one store will have no two records of the same item_number.
Now, I want to add some more custom conditions to my extract from this table; we are doing this manually on an xls file by applying some formula. Wondering if we can do this by a sql query.
Here are the conditions –
1.     From table a, for any particular assortment_number, count of items belonging to a store. E.g –
Count of items     Store_number
5     1011
7     1022
23     1033
12     1044
2.     This is more challenging. From table a, for stores with the same count of item_numbers, list down the store numbers which have the same count of items and same items in the set under them. E.g – as below.
Count of items     Store_number     Item_number     Match Indicator     Matching Stores
5     3     123|12|1234|12|1     N     
5     24     123|123|123|123|123     Y     22|234
5     21     123|451|2341|234|5100     N     
5     22     123|123|123|123|123     Y     24|234
5     456     123|12|1|2|3     N     
5     234     123|123|123|123|123     Y     22|24
3     45     12|123|12     Y     222|150
3     432     21|34|56     Y     123
3     222     12|123|12     Y     45|150
3     123     21|34|56     Y     432
3 150 12|123|12 Y 45|222
The above table may get positions shifted to its left due to formatting.
Appreciate your time and effort.
Thank You.

Answer 1

create table tmp_storeaggbyitem as
select
ltrim(trim(a.store_number),0) as store,
ltrim(trim(a.item_number),0) as item
from guide_pog_item_mstr a
where ltrim(trim(a.assortment_number),0) = '25571'
)Sample data in the table looks like this -
Store          Item
49               59310
49               59315
49               59114
49               59295
63               59310
63               59315
63               59314
63               59308
138             59310
138             59315
138             59114
138             59295
174             8399
174             6132
174             6129
174             6137
174             6175
270             8399
270             6132
270             6129
270             6137
270             6175
343             8399
343             6132
343             6129
343             6137
343             6175Here is the result set I am looking for -
Store        Count of Items        Item                                                   Matching Stores
49                    4                 59310|59315|59114|59295                    49|138    
63                    4                 59310|59315|59314|59308                    None
138                  4                 59310|59315|59114|59295                    49|138
174                  5                 8399|6132|6129|6137|6175                   174|270|343
270                  5                 8399|6132|6129|6137|6175                   174|270|343
343                  5                 8399|6132|6129|6137|6175                   174|270|343
.....Which version of the database are you using? The LISTAGG fucntuion (which is avaialble in Oracle 11.2, but not in 11.1) might be helpful here.
>
We are using Oracle Enterprise Edition 11.2.0.1.0 - 64 bit
Pl/Sql release 11.1.0.1.0
>
Why are you using the CHAR datatype instead of VARCHAR2? It may have nothing to do with this prolem, but it will cause other problems.
>
Agreed. I asked the same to my design team I inherited this table from. We are working out on the new table soon.
Thanks, Sanders.