Using Case in a select statement - sql

Consider the following table
create table temp (id int, attribute varchar(25), value varchar(25))
And values into the table
insert into temp select 100, 'First', 234
insert into temp select 100, 'Second', 512
insert into temp select 100, 'Third', 320
insert into temp select 101, 'Second', 512
insert into temp select 101, 'Third', 320
I have to deduce a column EndResult which is dependent on 'attribute' column. For each id, I have to parse through attribute values in the order
First, Second, Third and choose the very 1st value which is available i.e. for id = 100, EndResult should be 234 for the 1st three records.
Expected result:
| id | EndResult |
|-----|-----------|
| 100 | 234 |
| 100 | 234 |
| 100 | 234 |
| 101 | 512 |
| 101 | 512 |
I tried with the following query in vain:
select id, case when isnull(attribute,'') = 'First'
then value
when isnull(attribute,'') = 'Second'
then value
when isnull(attribute,'') = 'Third'
then value
else '' end as EndResult
from
temp
Result
| id | EndResult |
|-----|-----------|
| 100 | 234 |
| 100 | 512 |
| 100 | 320 |
| 101 | 512 |
| 101 | 320 |
Please suggest if there's a way to get the expected result.

You can use analytical function like dense_rank to generate a numbering, and then select those rows that have the number '1':
select
x.id,
x.attribute,
x.value
from
(select
t.id,
t.attribute,
t.value,
dense_rank() over (partition by t.id order by t.attribute) as priority
from
Temp t) x
where
x.priority = 1
In your case, you can conveniently order by t.attribute, since their alphabetical order happens to be the right order. In other situations you could convert the attribute to a number using a case, like:
order by
case t.attribute
when 'One' then 1
when 'Two' then 2
when 'Three' then 3
end

In case the attribute column have different values which are not in alphabetical order as is the case above you can write as:
with cte as
(
select id,
attribute,
value,
case attribute when 'First' then 1
when 'Second' then 2
when 'Third' then 3 end as seq_no
from temp
)
, cte2 as
(
select id,
attribute,
value,
row_number() over ( partition by id order by seq_no asc) as rownum
from cte
)
select T.id,C.value as EndResult
from temp T
join cte2 C on T.id = C.id and C.rownum = 1
DEMO

Here is how you can achieve this using ROW_NUMBER():
WITH t
AS (
SELECT *
,ROW_NUMBER() OVER (
PARTITION BY id ORDER BY (CASE attribute WHEN 'First' THEN 1
WHEN 'Second' THEN 2
WHEN 'Third' THEN 3
ELSE 0 END)
) rownum
FROM TEMP
)
SELECT id
,(
SELECT value
FROM t t1
WHERE t1.id = t.id
AND rownum = 1
) end_result
FROM t;
For testing purpose, please see SQL Fiddle demo here:
SQL Fiddle Example

keep it simple
;with cte as
(
select row_number() over (partition by id order by (select 1)) row_num, id, value
from temp
)
select t1.id, t2.value
from temp t1
left join cte t2
on t1.Id = t2.id
where t2.row_num = 1
Result
id value
100 234
100 234
100 234
101 512
101 512

Related

BigQuery - Recursive query to get leaf rows

I have a table in BigQuery which contains self referencing data like -
id | name | parent_id
----------------------------
1 | Gross Margin | null
2 | Revenue | 1
3 | Sales A | 2
4 | Sales B | 2
5 | 1001 | 3
6 | 1002 | 4
7 | OPEX | null
8 | Salaries | 7
9 | Payroll | 8
10 | Allowances | 9
11 | Commissions | 9
I want to write a query that returns the leaf rows of any row. For example if I give Gross Margin (or 1) as input, the query should return 1001 and 1002 (or 5 and 6) as output. Similarly if I give OPEX (or 7) as input, the query should return Allowances and Commissions (or 10 and 11) as output.
Updated script with improved convergence logic
DECLARE run_away_stop INT64 DEFAULT 0;
DECLARE flag BYTES;
CREATE TEMP TABLE ttt AS
SELECT parent_id, ARRAY_AGG(id order by id) children FROM `project.dataset.table` WHERE NOT parent_id IS NULL GROUP BY parent_id;
LOOP
SET (run_away_stop, flag) = (SELECT AS STRUCT run_away_stop + 1, md5(to_json_string(array_agg(t order by parent_id))) FROM ttt t);
CREATE OR REPLACE TEMP TABLE ttt1 AS
SELECT parent_id, ARRAY(SELECT DISTINCT id FROM UNNEST(children) id order by id) children
FROM (
SELECT parent_id, ARRAY_CONCAT_AGG(children) children
FROM (
SELECT t2.parent_id, ARRAY_CONCAT(t1.children, t2.children) children
FROM ttt t1, ttt t2
WHERE (SELECT COUNTIF(t1.parent_id = id) FROM UNNEST(t2.children) id) > 0
) GROUP BY parent_id
);
CREATE OR REPLACE TEMP TABLE ttt AS
SELECT * FROM ttt1 UNION ALL
SELECT * FROM ttt WHERE NOT parent_id IN (SELECT parent_id FROM ttt1);
IF (flag = (SELECT md5(to_json_string(array_agg(t order by parent_id))) FROM ttt t)) OR run_away_stop > 20 THEN BREAK; END IF;
END LOOP;
CREATE OR REPLACE TEMP TABLE ttt AS
SELECT id,
(
SELECT STRING_AGG(CAST(id AS STRING) order by id)
FROM ttt.children id
) children_as_list
FROM (SELECT DISTINCT id FROM `project.dataset.table`) d
LEFT JOIN ttt ON id = parent_id;
SELECT t1.id, STRING_AGG(child) leafs_list
FROM ttt t1, UNNEST(SPLIT(children_as_list)) child
JOIN (SELECT id FROM ttt WHERE children_as_list IS NULL) t2
ON t2.id = CAST(child AS INT64)
GROUP BY t1.id
ORDER BY id;
When applied to sample data in your question - it took 3 iterations and output is
Also, when applied to bigger example from your comments - it took 5 iterations and output is
BigQuery Team just introduced Recursive CTE! Hooray!!
With recursive cte you can use below approach
with recursive iterations as (
select id, parent_id from your_table
where not id in (
select parent_id from your_table
where not parent_id is null
)
union all
select b.id, a.parent_id
from your_table a join iterations b
on b.parent_id = a.id
)
select parent_id, string_agg('' || id order by id) as leafs_list
from iterations
where not parent_id is null
group by parent_id
If applied to sample data in your question - output is
Hope you agree it is more manageable and effective then when we were "forced" to use scripts for such logic!

Order rows by values

I trying order table by rank, but rows which have position value - have to have position according to value in position field. It is possible do it without additional tables, views etc?
I have table like this:
rank | position | name
999 | 10 | txt1
200 | 4 | txt2
32 | 1 | txt3
1200 | 2 | txt4
123 | null | txt5
234 | null | txt6
567 | null | txt7
234 | null | txt8
432 | null | txt9
877 | null | txt10
Desired output have to look like this:
rank | position | name
32 | 1 | txt3
1200 | 2 | txt4
877 | null | txt10
200 | 4 | txt2
567 | null | txt7
432 | null | txt9
345 | null | txt8
234 | null | txt6
123 | null | txt5
999 | 10 | txt1
Here is an idea. Assign the proper ordering to each row. Then, if the position is available use that instead. When there are ties, put the position value first:
select t.*
from (select t.*, row_number() over (order by rank desc) as seqnum
from t
) t
order by (case when position is not null then position else seqnum end),
(case when position is not null then 1 else 2 end);
SQL Fiddle doesn't seem to be working these days, but this query demonstrates the results:
with t(rank, position, t) as (
select 999, 10, 'txt1' union all
select 200, 4, 'txt2' union all
select 32 , 1, 'txt3' union all
select 1200, 2, 'txt4' union all
select 123, null, 'txt5' union all
select 234, null, 'txt6' union all
select 567, null, 'txt7' union all
select 234, null, 'txt8' union all
select 432, null, 'txt9' union all
select 877, null , 'txt10'
)
select t.*
from (select t.*, row_number() over (order by rank desc) as seqnum
from t
) t
order by (case when position is not null then position else seqnum end),
(case when position is not null then 1 else 2 end);
EDIT;
When I wrote the above, I had a nagging suspicion of a problem. Here is a solution that should work. It is more complicated but it does produce the right numbers:
with t(rank, position, t) as (
select 999, 10, 'txt1' union all
select 200, 4, 'txt2' union all
select 32 , 1, 'txt3' union all
select 1200, 2, 'txt4' union all
select 123, null, 'txt5' union all
select 234, null, 'txt6' union all
select 567, null, 'txt7' union all
select 234, null, 'txt8' union all
select 432, null, 'txt9' union all
select 877, null , 'txt10'
)
select *
from (select t.*, g.*,
row_number() over (partition by t.position order by t.rank) gnum
from generate_series(1, 10) g(n) left join
t
on t.position = g.n
) tg left join
(select t.*,
row_number() over (partition by t.position order by t.rank) as tnum
from t
) t
on tg.gnum = t.tnum and t.position is null
order by n;
This is a weird sort of interleaving problem. The idea is to create slots (using generate series) for the positions. Then, assign the known positions to the slots. Finally, enumerate the remaining slots and assign the values there.
Note: I hard-coded 10, but it is easy enough to put in count(*) from the table there.
suppose you stored data in table1.
Then you should update column "position" as follows:
update a
set position = x.pos_null
from table1
a
inner join
(
select
a.name,
COUNT(a.rank) as pos_null
from
(
select
*
from table1
where position is null
)
a
left join
(
select
*
from table1
)
b
on a.rank <= b.rank
group by
a.name
)
x
on a.name = x.name
select * from table1 order by position
Bye,
Angelo.

Difference between the values of multiple rows in SQL

My table in SQL is like:-
RN Name value1 value2 Timestamp
1 Mark 110 210 20160119
1 Mark 106 205 20160115
1 Mark 103 201 20160112
2 Steve 120 220 20151218
2 Steve 111 210 20151210
2 Steve 104 206 20151203
Desired Output:-
RN Name value1Lag1 value1lag2 value2lag1 value2lag2
1 Mark 4 3 5 4
2 Steve 9 7 10 4
The difference is calculated from the most recent to the second recent and then from second recent to the third recent for RN 1
value1lag1 = 110-106 =4
value1lag2 = 106-103 = 3
value2lag1 = 210-205 = 5
value2lag2 = 205-201 = 4
similarly for other RN's also.
Note: For each RN there are 3 and only 3 rows.
I have tried in several ways by taking help from similar posts but no luck.
I've assumed that RN and Name are linked here. It's a bit messy, but if each RN always has 3 values and you always want to check them in this order, then something like this should work.
SELECT
t1.Name
, AVG(CASE WHEN table_ranked.Rank = 1 THEN table_ranked.value1 ELSE NULL END) - AVG(CASE WHEN table_ranked.Rank = 2 THEN table_ranked.value1 ELSE NULL END) value1Lag1
, AVG(CASE WHEN table_ranked.Rank = 2 THEN table_ranked.value1 ELSE NULL END) - AVG(CASE WHEN table_ranked.Rank = 3 THEN table_ranked.value1 ELSE NULL END) value1Lag2
, AVG(CASE WHEN table_ranked.Rank = 1 THEN table_ranked.value2 ELSE NULL END) - AVG(CASE WHEN table_ranked.Rank = 2 THEN table_ranked.value2 ELSE NULL END) value2Lag1
, AVG(CASE WHEN table_ranked.Rank = 2 THEN table_ranked.value2 ELSE NULL END) - AVG(CASE WHEN table_ranked.Rank = 3 THEN table_ranked.value2 ELSE NULL END) value2Lag2
FROM table t1
INNER JOIN
(
SELECT
t1.Name
, t1.value1
, t1.value2
, COUNT(t2.TimeStamp) Rank
FROM table t1
INNER JOIN table t2
ON t2.name = t1.name
AND t1.TimeStamp <= t2.TimeStamp
GROUP BY t1.Name, t1.value1, t1.value2
) table_ranked
ON table_ranked.Name = t1.Name
GROUP BY t1.Name
There are other answers here, but I think your problem is calling for analytic functions, specifically LAG():
select
rn,
name,
-- calculate the differences
value1 - v1l1 value1lag1,
v1l1 - v1l2 value1lag2,
value2 - v2l1 value2lag1,
v2l1 - v2l2 value2lag2
from (
select
rn,
name,
value1,
value2,
timestamp,
-- these two are the values from the row before this one ordered by timestamp (ascending)
lag(value1) over(partition by rn, name order by timestamp asc) v1l1,
lag(value2) over(partition by rn, name order by timestamp asc) v2l1
-- these two are the values from two rows before this one ordered by timestamp (ascending)
lag(value1, 2) over(partition by rn, name order by timestamp asc) v1l2,
lag(value2, 2) over(partition by rn, name order by timestamp asc) v2l2
from (
select
1 rn, 'Mark' name, 110 value1, 210 value2, '20160119' timestamp
from dual
union all
select
1 rn, 'Mark' name, 106 value1, 205 value2, '20160115' timestamp
from dual
union all
select
1 rn, 'Mark' name, 103 value1, 201 value2, '20160112' timestamp
from dual
union all
select
2 rn, 'Steve' name, 120 value1, 220 value2, '20151218' timestamp
from dual
union all
select
2 rn, 'Steve' name, 111 value1, 210 value2, '20151210' timestamp
from dual
union all
select
2 rn, 'Steve' name, 104 value1, 206 value2, '20151203' timestamp
from dual
) data
)
where
-- return only the rows that have defined values
v1l1 is not null and
v1l2 is not null and
v2l1 is not null and
v2l1 is not null
This approach has the benefit that Oracle does all the necessary buffering internally, avoiding self-joins and the like. For big data sets this can be important from a performance viewpoint.
As an example, the explain plan for that query would be something like
-------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 6 | 150 | 13 (8)| 00:00:01 |
|* 1 | VIEW | | 6 | 150 | 13 (8)| 00:00:01 |
| 2 | WINDOW SORT | | 6 | 138 | 13 (8)| 00:00:01 |
| 3 | VIEW | | 6 | 138 | 12 (0)| 00:00:01 |
| 4 | UNION-ALL | | | | | |
| 5 | FAST DUAL | | 1 | | 2 (0)| 00:00:01 |
| 6 | FAST DUAL | | 1 | | 2 (0)| 00:00:01 |
| 7 | FAST DUAL | | 1 | | 2 (0)| 00:00:01 |
| 8 | FAST DUAL | | 1 | | 2 (0)| 00:00:01 |
| 9 | FAST DUAL | | 1 | | 2 (0)| 00:00:01 |
| 10 | FAST DUAL | | 1 | | 2 (0)| 00:00:01 |
-------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("V1L1" IS NOT NULL AND "V1L2" IS NOT NULL AND "V2L1" IS
Note that there are no joins, just a WINDOW SORT that buffers the necessary data from the "data source" (in our case, the VIEW 3 that is the UNION ALL of our SELECT ... FROM DUAL) to partition and calculate the different lags.
if just in this case, it's not that difficult.you need 2 steps
self join and get the result of minus
select t1.RN,
t1.Name,
t1.rm,
t2.value1-t1.value1 as value1,
t2.value2-t1.value2 as value2
from
(select RN,Name,value1,value2,
row_number(partition by Name order by Timestamp desc) as rm from table)t1
left join
(select RN,Name,value1,value2,
row_number(partition by Name order by Timestamp desc) as rm from table) t2
on t1.rm = t2.rm-1
where t2.RN is not null.
you set this as a table let's say table3.
2.you pivot it
select * from (
select t3.RN, t3.Name,t3.rm,t3.value1,t3.value2 from table3 t3
)
pivot
(
max(value1)
for rm in ('1','2')
)v1
3.you get 2 pivot table for value1 and value2 join them together to get the result.
but i think there may be a better way and i m not sure if we can just join pivot when we pivot it so i ll use join after i get the pivot result that will make 2 more tables. its not good but the best i can do
-- test data
with data(rn,
name,
value1,
value2,
timestamp) as
(select 1, 'Mark', 110, 210, to_date('20160119', 'YYYYMMDD')
from dual
union all
select 1, 'Mark', 106, 205, to_date('20160115', 'YYYYMMDD')
from dual
union all
select 1, 'Mark', 103, 201, to_date('20160112', 'YYYYMMDD')
from dual
union all
select 2, 'Steve', 120, 220, to_date('20151218', 'YYYYMMDD')
from dual
union all
select 2, 'Steve', 111, 210, to_date('20151210', 'YYYYMMDD')
from dual
union all
select 2, 'Steve', 104, 206, to_date('20151203', 'YYYYMMDD') from dual),
-- first transform value1, value2 to value_id (1,2), value
data2 as
(select d.rn, d.name, 1 as val_id, d.value1 as value, d.timestamp
from data d
union all
select d.rn, d.name, 2 as val_id, d.value2 as value, d.timestamp
from data d)
select * -- find previous row P of row D, evaluate difference and build column name as desired
from (select d.rn,
d.name,
d.value - p.value as value,
'value' || d.val_id || 'Lag' || row_number() over(partition by d.rn, d.val_id order by d.timestamp desc) as col
from data2 p, data2 d
where p.rn = d.rn
and p.val_id = d.val_id
and p.timestamp =
(select max(pp.timestamp)
from data2 pp
where pp.rn = p.rn
and pp.val_id = p.val_id
and pp.timestamp < d.timestamp))
-- pivot
pivot(sum(value) for col in('value1Lag1',
'value1Lag2',
'value2Lag1',
'value2Lag2'));

How to comapre two columns of a table in sql?

In a table there are two columns:
-----------
| A | B |
-----------
| 1 | 5 |
| 2 | 1 |
| 3 | 2 |
| 4 | 1 |
-----------
Want a table where if A=B then
-------------------
|Match | notMatch|
-------------------
| 1 | 5 |
| 2 | 3 |
| Null | 4 |
-------------------
How can i do this?
I tried something which shows the Matched part
select distinct C.A as A from Table c inner join Table d on c.A=d.B
Try this:
;WITH TempTable(A, B) AS(
SELECT 1, 5 UNION ALL
SELECT 2, 1 UNION ALL
SELECT 3, 2 UNION ALL
SELECT 4, 1
)
,CTE(Val) AS(
SELECT A FROM TempTable UNION ALL
SELECT B FROM TempTable
)
,Match AS(
SELECT
Rn = ROW_NUMBER() OVER(ORDER BY Val),
Val
FROM CTE c
GROUP BY Val
HAVING COUNT(Val) > 1
)
,NotMatch AS(
SELECT
Rn = ROW_NUMBER() OVER(ORDER BY Val),
Val
FROM CTE c
GROUP BY Val
HAVING COUNT(Val) = 1
)
SELECT
Match = m.Val,
NotMatch= n.Val
FROM Match m
FULL JOIN NotMatch n
ON n.Rn = m.Rn
Try with EXCEPT, MINUS and INTERSECT Statements.
like this:
SELECT A FROM TABLE1 INTERSECT SELECT B FROM TABLE1;
You might want this:
SELECT DISTINCT
C.A as A
FROM
Table c
LEFT OUTER JOIN
Table d
ON
c.A=d.B
WHERE
d.ID IS NULL
Please Note that I use d.ID as an example because I don't see your schema. An alternate is to explicitly state all d.columns IS NULL in WHERE clause.
Your requirement is kind of - let's call it - interesting. Here is a way to solve it using pivot. Personally I would have chosen a different table structure and another way to select data:
Test data:
DECLARE #t table(A TINYINT, B TINYINT)
INSERT #t values
(1,5),(2,1),
(3,2),(4,1)
Query:
;WITH B AS
(
( SELECT A FROM #t
EXCEPT
SELECT B FROM #t)
UNION ALL
( SELECT B FROM #t
EXCEPT
SELECT A FROM #t)
), A AS
(
SELECT A val
FROM #t
INTERSECT
SELECT B
FROM #t
), combine as
(
SELECT val, 'A' col, row_number() over (order by (select 1)) rn FROM A
UNION ALL
SELECT A, 'B' col, row_number() over (order by (select 1)) rn
FROM B
)
SELECT [A], [B]
FROM combine
PIVOT (MAX(val) FOR [col] IN ([A], [B])) AS pvt
Result:
A B
1 3
2 4
NULL 5

tSQL UNPIVOT of comma concatenated column into multiple rows

I have a table that has a value column. The value could be one value or it could be multiple values separated with a comma:
id | assess_id | question_key | item_value
---+-----------+--------------+-----------
1 | 859 | Cust_A_1 | 1,5
2 | 859 | Cust_B_1 | 2
I need to unpivot the data based on the item_value to look like this:
id | assess_id | question_key | item_value
---+-----------+--------------+-----------
1 | 859 | Cust_A_1 | 1
1 | 859 | Cust_A_1 | 5
2 | 859 | Cust_B_1 | 2
How does one do that in tSQL on SQL Server 2012?
We have a user defined function that we use for stuff like this that we called "split_delimiter":
CREATE FUNCTION [dbo].[split_delimiter](#delimited_string VARCHAR(8000), #delimiter_type CHAR(1))
RETURNS TABLE AS
RETURN
WITH cte10(num) AS
(
SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL
SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL
SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1
)
,cte100(num) AS
(
SELECT 1
FROM cte10 t1, cte10 t2
)
,cte10000(num) AS
(
SELECT 1
FROM cte100 t1, cte100 t2
)
,cte1(num) AS
(
SELECT TOP (ISNULL(DATALENGTH(#delimited_string),0)) ROW_NUMBER() OVER (ORDER BY (SELECT NULL))
FROM cte10000
)
,cte2(num) AS
(
SELECT 1
UNION ALL
SELECT t.num+1
FROM cte1 t
WHERE SUBSTRING(#delimited_string,t.num,1) = #delimiter_type
)
,cte3(num,[len]) AS
(
SELECT t.num
,ISNULL(NULLIF(CHARINDEX(#delimiter_type,#delimited_string,t.num),0)-t.num,8000)
FROM cte2 t
)
SELECT delimited_item_num = ROW_NUMBER() OVER(ORDER BY t.num)
,delimited_value = SUBSTRING(#delimited_string, t.num, t.[len])
FROM cte3 t;
GO
It will take a varchar value up to 8000 characters and will return a table with the delimited elements broken into rows. In your example, you'll want to use an outer apply to turn those delimited values into separate rows:
SELECT my_table.id, my_table.assess_id, question_key, my_table.delimited_items.item_value
FROM my_table
OUTER APPLY(
SELECT delimited_value AS item_value
FROM my_database.dbo.split_delimiter(my_table.item_value, ',')
) AS delimited_items