SQL - Using Row Number Over Partition By - With Grouping Sets
Jan 2, 2013
I am trying to write a single SELECT statement that groups at 2 levels of aggregation (using grouping sets) and assigns row numbers (to rank each item) that are partitioned at the correct level for each grouping set. I have the grouping sets figured out but I can't find a way to make Partition By match each level of aggregation.
What I am looking for (in a single SELECT statement) is logically equivalent to:
SELECTweek
,region
,NULL as country
,item
,SUM(qty)
,ROW_NUMBER OVER (PARTITION BY week, region ORDER BY SUM(qty) DESC) as rownumFROM base
GROUP BY week
,region
,item
UNION ALL
SELECTweek
,NULL as region
,country
,item
,SUM(qty)
,ROW_NUMBER OVER (PARTITION BY week, country ORDER BY SUM(qty) DESC) as rownumFROM base
GROUP BY week
,country
,item
I hoped that I could do something like this:
SELECTweek
,region
,country
,item
,SUM(qty)
,ROW_NUMBER OVER (PARTITION BY week, GROUPING SETS (region, country) ORDER BY SUM(qty) DESC) as rownumFROM base
GROUP BY week
,GROUPING SETS (region, country)
,item
But it looks like I am not allowed to partition by grouping sets -- I get the error ORA-00907: missing right parenthesis. I didn't expect it to work but I am not sure how else to partition by multiple levels.
let me know if I could have tagged my code or met other forum standards better.
The below sql is giving different number of result sets while adding further columns in select clause.i.e After adding the columns 4,5,6 in the below query its giving different number of result set.In this case the result set count would be 5.
Before adding the columns 4,5,6,the result set count was 11.
For the above data set i need to divide into 5 partittions and need to updated the partitonid with the partition number for each partition set,like the below result set
If I run the following query I got 997 records by using GROUP BY.
SELECT c.ins_no, b.pd_date,a.project_id, a.tech_no FROM mis.tranche_balance a, FMSRPT.fund_reporting_period b, ods.proj_info_lookup c, ods.institution d WHERE a.su_date = b.pd_date AND a.project_id = c.project_id AND c.ins_no = d.ins_no AND d.sif_code LIKE 'P%' AND d.sif_code <> 'P-DA' AND a.date_stamp >='01-JAN-2011' AND pd_date='31-MAR-2011' GROUP BY c.ins_no, b.pd_date, a.project_id, a.tech_no;
I want to show the extra columns a.date_stamp and a.su_date in the out put so that I have used PARTITION BY in the second query but I got 1079 records.
i have table with range partition and list sub-partition..can i add one more list sub-partition if it is not possible , i have to drop first sub-partition.
I need to populate a table based on the results of comparing sets of data. I decided to do this using MULTISET EXCEPT, but having created the structure, do not know whether it is actually possible, and if so, what syntax to use.
I have created:
CREATE OR REPLACE TYPE NUMBER_TBL IS TABLE OF NUMBER; /
CREATE OR REPLACE TYPE PACKAGE_OPTION_GROUP_OBJ AS OBJECT( ID NUMBER, benefits NUMBER_TBL ) /
CREATE OR REPLACE TYPE PACKAGE_OPTION_GROUP_TBL AS table of PACKAGE_OPTION_GROUP_OBJ /
I am expecting the result to be something like: ***comparisonTable*** GROUP_ID : 123 BENEFITS : 165 BENEFITS : 167 ---------------------------
However, I don't know if this is possible and if so, what the syntax would be.
At a later stage, I will need to compare the benefits between selectedTable and groupTable, where the GroupID's match, which is why I have the tables structured in this way.
I have two queries. One selects 5 values (all number datatypes), the second query selects 2 values (both number datatypes).If a value from query 2 exists in the resultset of query 1 - go make table updates using the second value in the select statment of query 2. If there is not a match, update a table with values from query 1.
Currently, I have two cursor for loops, which are nested. This works but of course is horribly slow. What is a better way to process this data?The number of rows varies between the two resultsets.
have two queries that will return same columns from two different set of tables ( column mapping has been taken care of). The return type is out ref cursor. (P_SUPPLY_REORDER )
Query-2 Xcom -------------------------------------- select null as sMO_NO, xso.created_date as SPLY_ORD_DT, xso.fk_cust_id as cust_id, cust.cust_po_no as cust_PO_NO ,(sta.SHIP_TO_ATTN_FIRST_NAME||''||sta.SHIP_TO_ATTN_LAST_NAME) as attention_name, xsol.CARTONS_ORDERED as SPLY_ORD_QTY, [code].......
Now the requirement is One of four conditions are possible for each Supply Reorder Number:
. Both table queries return no records- Populate all the P_SUPPLY_REORDER output fields with nulls . SUPPLY_ORDER returns a record, but XCOM_ORDER_HEADER returns no records - Populate output fields with values from the join of SUPPLY_ORDER and SUPPLY_ORDER_LINE. . SUPPLY_ORDER returns no records, but XCOM_ORDER_HEADER returns one record - Populate output fields with values from the join of XCOM_ORDER_HEADER and XCOM_ORDER_LINES. . SUPPLY_ORDER returns a record, and XCOM_ORDER_HEADER returns a record; find out the latest order by comapring max(SPLY_ORD_DT) from SUPPLY_ORDER with max(CREATED_DATE) from XCOM_ORDER_HEADER. - If the latest order is in SUPPLY_ORDER, then populate output fields with values from the join of SUPPLY_ORDER and SUPPLY_ORDER_LINE. - If order dates are equal from both join results, then populate output fields with values from the join of SUPPLY_ORDER and SUPPLY_ORDER_LINE. - If the latest order is in XCOM_ORDER_HEADER, then populate output fields with values from the join of XCOM_ORDER_HEADER and XCOM_ORDER_LINES.
Question is how can we switch over the queries to pull respective dataset based on these conditions ( checking that which table join is going to return a row and then based upon latest order if both tables return a row) and all this logic as part of single SQL statement that is returned as OUT Ref Cursor.
i basically need to setup a schema for globalization. i need to handle brittish dollar signs, american dollar signs. etc.what is a global character setting i can use to handle almost everything?
my problem right now is that when i store a brittian dollar sign it converts it to a box. i need this preserved so i can select it and use it later.will the WE8ISO8859P1 character setting work? where is a good place to view all of the types of character settings? and what they are for. etc. and also how do you change such things? not for the session, but for a schema wide
here are my current sessions
NLS_LANGUAGE AMERICAN NLS_TERRITORY AMERICA NLS_CURRENCY $ NLS_ISO_CURRENCY AMERICA
I know how to select the last N sets of rows, using DENSE_RANK - where multiple rows have the same timestamp but I want to only select those rows which do NOT have the top 2 unique timestamps.
i.e.:
SELECT * FROM ( SELECT DENSE_RANK() OVER (ORDER BY myTimestamp DESC) DENSE_RANK, HISTORYID, USER_ID, myTimestamp, STATUS, FROM TXN_HIST) WHERE DENSE_RANK > 2 order by myTimestamp DESC, HISTORYID, USER_ID;
I Know we can create dynamic partitions on table in oracle 11g. Is it possible to create normal partition and sub partition both dynamically.I have to create Normal partition range on date and sub partition list on Batch ID (varchar).
I have a table that partitioned into six partitions. each partitions placed in different table space and every two table space placed it on a different hardisk
when I will do query select with the non-partition keys condition, how the search process ? whether the sequence (scan sequentially from partition 1 to partition 6) or partition in a hardisk is accessed at the same time with other partition in other hardisk. ( in the image, partition 1,4 accessed at the same time with partition 2,5 and 3,6)
Can I add range sub partition to a hash partition table. Example like this.
CREATE TABLE test ( test_id VARCHAR2(10 ) , test_TYPE VARCHAR2(5) , CREATE_DATE date ) partition by hash (test_id, test_type) Partitions 3 SUBPARTITION BY RANGE (CREATE_DATE);
When Tried, I am getting syntax error as invalid option.
Activity Structure ACTIVITY_TYPE CHAR (1) NOT NULL, ACTIVITY_DATE DATE DEFAULT sysdate NOT NULL, ACTIVITY_ON VARCHAR2 (30) NOT NULL, REFERENCE_NO VARCHAR2 (19), CHILD_REFERENCE_NO VARCHAR2 (19), USER_ID VARCHAR2 (30) DEFAULT user NOT NULL, TERMINAL VARCHAR2 (30) DEFAULT userenv ('TERMINAL') NOT NULL )
Activity1 Structure Which I have Done Partitioning When I insert data from Activity to Activity1 it gives that error ORA-14400: inserted partition key does not map to any partition what I am doing wrong
CREATE TABLE ACTIVITY1( ACTIVITY_TYPE CHAR (1) NOT NULL, ACTIVITY_DATE DATE DEFAULT sysdate NOT NULL, ACTIVITY_ON VARCHAR2 (30) NOT NULL, REFERENCE_NO VARCHAR2 (19),
[code]....
Insert Statement
insert into ACTIVITY1(ACTIVITY_TYPE, ACTIVITY_DATE, ACTIVITY_ON, REFERENCE_NO, CHILD_REFERENCE_NO,
I have partition based table one the basis of year month. And we have 8 local indexes on this table. Every month we have to create a new partition and load data into this partition and the volume of the data is around 14million and the load process is taking long time due to indexes. Is it possible to drop the indexes from particular partition?
When I am trying to insert record from tbl_mittal into tbl_temp table. I am facing "ORA-14400: inserted partition key does not map to any partition" error
SQL> insert into tbl_temp select * from tbl_mittal; insert into tbl_temp select * from tbl_mittal * ERROR at line 1: ORA-14400: inserted partition key does not map to any partition
AS tbl_mittal is having hugh number of records so I am providing only few rows from tbl_mittal table as test data.
I have a requirement to get the records group wise.Ex: For each departments, i need to get the employee details as a coma seperated.It means that the output must have the department name in first column and the second column must contain all the employees in that particular department (As a coma seperated).
In the below data, a container is moving from one city to another. 1,2 ,3 can be any number which i want to generate and use as keys to group the cities. Eg: AUH, JEB, CIW belong to the same key=2; SIN, IKT belong to a new group 4. The City where difference between the Seq# is greater than 1 (eg between S8W and AUH), a new group starts.
Name Hours date a810/11/2011 a 510/12/2011 a610/13/2011 a710/14/2011 a710/15/2011 a810/16/2011 a710/17/2011 a810/18/2011 a810/19/2011 a710/20/2011 a710/21/2011
If i want the sum of hours for 3 days range ,how should i do it.
E.g. say
name hrs startdate enddate
a 19 10/11/2011 10/13/2011 a 22 10/14/201110/16/2011 a 23 10/17/2011 10/19/2011
How can I get the grp_id for unique combination of manager and department, grp_id should be created on asc order of manager_id. In this example manager_id 100 is minimum, so it should be grp 1 and all the employees with that manager_id should be in grp_id 1, for manager_id 114 grp_id should be 2.
If, there is manager_id 117, it should create grp_id 3.
To get grp_num ,I can use row_number() over (partition by department_id,manager_id order by employee_id) grp_num
I am looking for an update statement for this issue.
And the problem is, that when i use sutp_price_proc and pbk_price in grouping, it splits my results by those rows. If i delete them from grouping, sql gives me error about not a single grouping in line 1.
pas_codepas_profilesutp_idsutp_pricex 2664good stuff310069< because pbk_price is like 67 from that period 2664good stuff310071< because pbk_price is like 50 from other period
