Question related to Grouping

Question

Dear Members,
My Data Looks like :
Period Name *** Description ***** Amount
Dec-07 ******* TEST123     ******250
Dec-07 ******* TEST456     ******560
Dec-07 ******* TEST456     ******500
Dec-07 ******* TEST789     ******750
Jan-08 ******* SALES     ******670
Jan-08 ******* TRIP     ******860
Jan-08 ******* TEST123     ******650
I have 3 columns Period Name, description,amount. I have to write a sql query which shows my data as below:
Description********* DEC-07 **************** JAN-08
TEST123 ********** 250 *******************650
TEST456 ********** 1060 *******************0
TEST789 ********** 750 *******************0
SALES ************ 0 *********************670
TRIP **************0 *********************860
Basically i need amt for all descriptions monthly. for example sales exsists only in jan-08 so amount will be 0 in dec-07 and will be 860 in jan-08.
It should also group all descriptions which have same name under a particular period. for example TEST456 repeats twice in dec-07 period so we display 1060 under dec-07 and 0 under jan-08
How can we achieve this.
Thanks

Answer 1

Hi,
Sorry, I don't understand what you want.
String aggregation (in this context) means producing one big string that is formatted so that it looks like multiple columns. Exactly how many columns, and what each one represents, can be determined by the query itself.
The output from a string aggregation query might be:
DESCRIPTION AGG_DATA
              Dec-07  Jan-08
SALES              0     670
TEST123          250     650
TEST456         1060       0
TEST789          750       0
TRIP               0     860Note that there are two columns of output:
(1) description
(2) agg_data, which is formatted to look like two columns, but is actually just one string. With different data, it could look like 3, or 4, or 34 columns. The code producing it would be not change in the slightest.
Also not e that there are 6 rows of output: one that looks like column headers, and 5 (one for each description) wth data.
A completely different (and harder) approach is Dynamic SQL, where you run some preliminary queries (and maybe some PL/SQL code) to determijne how many columns and what each column represents. Using this information, you then:
(1) produce a "main" query that has exactly as many columns as needed for the data, and
(2) run that main query
With dynamic SQL you really would get separate columns for each month. Since the number of columns in a query must be hard-coded into the query, you don't write the main query now; you write the preliminary queries that will write a main query.
Have you decided which approach
spool you want to try?