Group by and then convert form Row to column

Question

Hi All,
below is a transaction table for bank customer.
region_id     cutomer_id     transaction_type     transaction_date
101          12345          CC               10-March-2011
101          12345          DC               07-March-2011
101          12345          P1               01-March-2011
101          12345          p2               10-Jan-2011
102          45678          CC               15-feb-2011
101          12345          p3               27-OCT-2010with combinaton of region_id and customer_id, we get uniq record from the table. There are also different kind of transaction a customer has done. It is given in transaction_type column (CC - Credit card, Dc - Debit Card, and those start with P% {including P1, p2, p3} = payment).
Now my requirement is to get the number of transaction for each transaction type for a current year and also number of total transaction by the customer. Below is my output table.
region_id     cutomer_id     transaction_type     transaction_date     Tran_type_cc     Tran_type_DC     Tran_type_P     count_tran
101          12345          CC               10-March-2011          1          1          2          5
101          12345          DC               07-March-2011          1          1          2          5
101          12345          P1               01-March-2011          1          1          2          5
101          12345          p2               10-Jan-2011          1          1          2          5
102          45678          CC               15-feb-2011          1          0          0          1
101          12345          p3               27-OCT-2010          1          1          2          5if i do group by it gives me result but at row level and to convert row into column to get above result, i need to use decode function. My sourcec is containing 25 miliion records, and if i execute the query, it is getting hanged. Can someone help me how can i write a fine tuned query by using analytical function

Answer 1

How about this?
Note: I suppose, p1, P1, p2, P2, p3, P3 all belong to the same payment category regardless of their upper/lower case.
Here is the table and the data:
CREATE TABLE T1 (REGION_ID NUMBER, CUSTOMER_ID NUMBER, TRAN_TYPE VARCHAR2(5), TRAN_DATE DATE);
INSERT INTO T1 VALUES (101, 12345, 'CC', TO_DATE('10-Mar-2011','DD-MON-YYYY'));
INSERT INTO T1 VALUES (101, 12345, 'DC', TO_DATE('07-Mar-2011','DD-MON-YYYY'));
INSERT INTO T1 VALUES (101, 12345, 'P1', TO_DATE('01-Mar-2011','DD-MON-YYYY'));
INSERT INTO T1 VALUES (101, 12345, 'p2', TO_DATE('10-Jan-2011','DD-MON-YYYY'));
INSERT INTO T1 VALUES (102, 45678, 'CC', TO_DATE('15-FEB-2011','DD-MON-YYYY'));
INSERT INTO T1 VALUES (101, 12345, 'p3', TO_DATE('27-OCT-2010','DD-MON-YYYY'));Here is the query:
SELECT REGION_ID, CUSTOMER_ID, TRAN_TYPE, TRAN_DATE
  , SUM(DECODE(TRAN_TYPE, 'CC', '1','0')) OVER (PARTITION BY CUSTOMER_ID) AS CC
  , SUM(DECODE(TRAN_TYPE, 'DC', '1', '0')) OVER (PARTITION BY CUSTOMER_ID ) AS DC
  , SUM(DECODE(SUBSTR(UPPER(TRAN_TYPE), 1,1) , 'P', '1','0')) OVER (PARTITION BY CUSTOMER_ID) AS P
  , SUM(1) OVER (PARTITION BY CUSTOMER_ID) AS TRAN_T
FROM T1 ;
r_id   cust_id      t_TYPE   T_DATE  CC_CNT  DC_CNT  P_CNT  T_CNT
101     12345     CC     10-MAR-11     1     1     3     5
101     12345     DC     07-MAR-11     1     1     3     5
101     12345     p3     27-OCT-10     1     1     3     5
101     12345     p2     10-JAN-11     1     1     3     5
101     12345     P1     01-MAR-11     1     1     3     5
102     45678     CC     15-FEB-11     1     0     0     1Edited by: PhoenixBai on Mar 14, 2011 4:16 PM