How i achive this data by writing exact query ?

Question

Dear Exparts,
Oracle Database 10g Enterprise Edition Release 10.2.0.3.0 - Prod
PL/SQL Release 10.2.0.3.0 - Production
"CORE     10.2.0.3.0     Production"
TNS for 32-bit Windows: Version 10.2.0.3.0 - Production
NLSRTL Version 10.2.0.3.0 - ProductionI have a table Name TB and Column A and B with the following data.
A       B
10     02-AUG-2012
10     04-AUG-2012
10     01-SEP-2012
10     10-AUG-2012
11     04-AUG-2012
11     01-SEP-2012
11     01-AUG-2012
10     12-AUG-2012
10     15-JUL-2012I need data like
A                      B
10            10-AUG-2012
10            12-AUG-2012
10            01-SEP-2012
11            01-SEP-2012How we get this data ?
We have to count column B group by Column A. then devided the count_value by 2 and trunc it and base on the result out put will come. the data are the maximum data.
Like when we count group A result will be
A            B
10           3
11           1base on above result my desire outpul will come...
Here is the script
CREATE TABLE TB
    "A" VARCHAR2(20 BYTE),
    "B" DATE
Insert into TB (A,B) values ('10',to_date('02-AUG-2012','DD-MON-RRRR'));
Insert into TB (A,B) values ('10',to_date('04-AUG-2012','DD-MON-RRRR'));
Insert into TB (A,B) values ('10',to_date('01-SEP-2012','DD-MON-RRRR'));
Insert into TB (A,B) values ('10',to_date('10-AUG-2012','DD-MON-RRRR'));
Insert into TB (A,B) values ('11',to_date('04-AUG-2012','DD-MON-RRRR'));
Insert into TB (A,B) values ('11',to_date('01-SEP-2012','DD-MON-RRRR'));
Insert into TB (A,B) values ('11',to_date('01-AUG-2012','DD-MON-RRRR'));
Insert into TB (A,B) values ('10',to_date('12-AUG-2012','DD-MON-RRRR'));
Insert into TB (A,B) values ('10',to_date('15-JUL-2012','DD-MON-RRRR'));Thanks in advance...
Ask2Learn
Edited by: Asked to Learn on Sep 1, 2012 7:55 AM
Edited by: Asked to Learn on Sep 1, 2012 8:38 AM (CHANGE DATA TYPE) SORRY

Answer 1

this ?
with t as
select a,b,round(count(*) over(partition by a)/2) cnt,row_number() over(order by null) rn from tb
),t1 as
select a.*,row_number() over(order by null) rn from tb a
select t1.a,t1.b from t,t1 where t1.rn=t.rn and  t1.b>t.cnt order by a;
A                    B                     
10                   4                     
10                   6                     
10                   5                     
11                   3                     
4 rows selected
Or
with t as
select a,b,round(count(*) over(partition by a)/2) cnt from tb
select a,b from t where cnt<b;