Reset sum when condition is met in Oracle - sql

My data is structured as follows:
Timestamp | Hour | Count
--------------------------
20190801 01 | 1 | 10
20190801 02 | 2 | 20
20190801 03 | 3 | 10
20190801 04 | 4 | 5
20190801 05 | 5 | 15
20190801 06 | 6 | 10
20190802 01 | 1 | 5
20190802 02 | 2 | 20
20190802 03 | 3 | 5
20190802 04 | 4 | 15
20190802 05 | 5 | 20
20190802 06 | 6 | 5
20190803 01 | 1 | 30
I'm trying to make an SQL query that will calculate a running SUM but resets when the hour is 3. The result should look like this:
Hour | Count | SUM
------------------
1 | 10 | 10
2 | 20 | 30
3 | 10 | 10 /* RESET */
4 | 5 | 15
5 | 15 | 30
6 | 10 | 40
1 | 5 | 45
2 | 20 | 65
3 | 5 | 5 /* RESET */
4 | 15 | 20
5 | 20 | 40
6 | 5 | 45
1 | 30 | 75


You could create subgroup using conditional sum:
WITH cte AS (
SELECT t.*,SUM(CASE WHEN hour=3 THEN 1 ELSE 0 END) OVER(ORDER BY timestamp) grp
FROM t
)
SELECT cte.*, SUM(Count) OVER(PARTITION BY grp ORDER BY timestamp) AS total
FROM cte

Related

Integrate row information with previous rows SQL

I need to integrate row information with previous rows
| ID | no| number |
+--------------+-----+--------+
| 1 | 40| 10 |
| 2 | 32| 12 |
| 3 | 40| 15 |
| 4 | 45| 23 |
| 5 | 32| 15 |
| 6 | 12| 14
| 7 | 40| 20
| 8 | 32| 18
| 9 | 45| 27
| 10 | 12| 16
Desired result :
| ID | no | number | last number
+--------------+-----+--------+-------------
| 1 | 40 | 10 | 0
| 3 | 32 | 12 | 0
| 3 | 40 | 15 | 0
| 4 | 45 | 23 | 0
| 5 | 32 | 15 | 12
| 6 | 12 | 14 | 0
| 7 | 40 | 20 | 15
| 8 | 32 | 18 | 15
| 9 | 45 | 27 | 23
| 10 | 12 | 16 | 14

The best guess from me is - you are looking for a script as below. But according to the below logic, row with "id = 3" should get 10 as value in the column 'last number'
You can check the DEMO HERE
SELECT *,
ISNULL
(
(
SELECT number
FROM your_table C
WHERE C.ID =
(
SELECT MAX(ID) FROM your_table B WHERE B.ID < A.ID AND B.no = A.No
)
)
,0) [last number]
FROM your_table A
Output is-
ID no number last number
1 40 10 0
2 32 12 0
3 40 15 10
4 45 23 0
5 32 15 12
6 12 14 0
7 40 20 15
8 32 18 15
9 45 27 23
10 12 16 14

Checking for Consecutive 12 Weeks of 0 Sales

I have a table with customer_number, week, and sales. I need to check if there were 12 consecutive weeks of no sales for each customer and create a flag of 0/1.
I can check the last 12 weeks or a certain time frame, but what's the best way to check for consecutive runs? Here is the code I have so far:
select * from weekly_sales
where customer_nbr in (123, 234)
and week < '2015-11-01'
and week > '2014-11-01'
order by customer_nbr, week
;

Sql Fiddle Demo
Here is a simplify version only need a week_id and sales
SELECT S1.weekid start_week, MAX(S2.weekid) end_week, SUM (S2.sales)
FROM Sales S1
JOIN Sales S2
ON S2.weekid BETWEEN S1.weekid and S1.weekid + 11
WHERE S1.weekid BETWEEN 1 and 25 -- your search range
GROUP BY S1.weekid
Let me know if that work for you
OUTPUT
| start_week | end_week | |
|------------|----------|----|
| 1 | 12 | 12 |
| 2 | 13 | 8 |
| 3 | 14 | 3 |
| 4 | 15 | 2 |
| 5 | 16 | 0 | <-
| 6 | 17 | 0 | <- no sales for 12 week
| 7 | 18 | 0 | <-
| 8 | 19 | 4 |
| 9 | 20 | 9 |
| 10 | 21 | 11 |
| 11 | 22 | 15 |
| 12 | 23 | 71 |
| 13 | 24 | 78 |
| 14 | 25 | 86 |
| 15 | 25 | 86 | < - less than 12 week range
| 16 | 25 | 86 | < - below this line
| 17 | 25 | 86 |
| 18 | 25 | 86 |
| 19 | 25 | 86 |
| 20 | 25 | 82 |
| 21 | 25 | 77 |
| 22 | 25 | 75 |
| 23 | 25 | 71 |
| 24 | 25 | 15 |
| 25 | 25 | 8 |
Your final query should have
HAVING SUM (S2.sales) = 0
AND COUNT(*) = 12

Ummmmm...You could use between 'week' and 'week', and you can use too the "count(column)" in order to improve performance.
So you only have to compare if result is bigger than 0

Complex SQL query with pivot

I have the following table.
Data_table
R_id I_id Metric CType Timespan Quantity Date
1 1 S C Week 100 4/5/2015
1 1 Q C Week 200 4/5/2015
1 1 I D Week 80 4/5/2015
1 2 S C Week 150 4/5/2015
1 2 Q C Week 100 4/5/2015
1 2 I D Week 50 4/5/2015
Metric can have a limited set of values (S, Q, I..)
CType will be C, D or nil.
Timespan can be Weekly/Daily.
Date will be a Sunday (start of week) for Weekly and that day's date for Daily.
My goal is to convert this to a daily view which would involve
If Timespan is Daily, copy the Quantity for the above metrics as it is.
Converting a Weekly quantity to 7 Daily quantities.
If the CType is D copy the quantity as it is.
If the CType is C use a constant percentage breakdown logic to distribute the weekly over 7 days.eg [30%, 10%, 10%, 5%, 10%, 15% 20%] = 100%
Creating the following VIEW.
R_id I_id Date S Q I ... (other metrics whose CType is not nil)
1 1 4/5/2015 30 60 80 ... (the quantity of the other metrics)
1 1 4/6/2015 10 20 80
1 1 4/7/2015 10 20 80
1 1 4/8/2015 5 10 80
1 1 4/9/2015 10 20 80
1 1 4/10/2015 15 30 80
1 1 4/11/2015 20 40 80
1 2 4/5/2015 45 30 50
1 2 4/6/2015 15 10 50
1 2 4/7/2015 15 10 50
1 2 4/8/2015 7.5 5 50
1 2 4/9/2015 15 10 50
1 2 4/10/2015 22.5 15 50
1 2 4/11/2015 30 20 50
I can write a bunch of java methods which will pull out the data from the above table and get the values for metrics as needed. But for a large dataset, the performance will not be very good. Databases are meant for this type of data computation. Once this view is created, I can quickly (and simply) query it to get what I want. I can write simple sql queries. But I have no clue how to even begin approaching this problem! I can see a PIVOT here (logically, I don't know how a query would or even can achieve it). But how to compute the 7 daily quantities from a weekly quantity and put it in the VIEW?
Suggestions and guidance will be much appreciated.

You can use hierarchical queries to generate daily data.
SQL Fiddle
Query:
select
r_id,
i_id,
metric,
ctype,
timespan,
quantity,
tdate + level - 1 as m_tdate,
level as m_level,
(case ctype
when 'C' then
(case level
when 1 then 0.3
when 2 then 0.1
when 3 then 0.1
when 4 then 0.05
when 5 then 0.1
when 6 then 0.15
when 7 then 0.2
end)
else 1
end) * quantity as m_quantity
from myt
where timespan = 'Week'
connect by level <= 7
and r_id = prior r_id
and i_id = prior i_id
and metric = prior metric
and ctype = prior ctype
and timespan = prior timespan
and prior sys_guid() is not null
This will generate seven day data for each record
Results:
| R_ID | I_ID | METRIC | CTYPE | TIMESPAN | QUANTITY | M_TDATE | M_LEVEL | M_QUANTITY |
|------|------|--------|-------|----------|----------|-----------------------|---------|------------|
| 1 | 1 | I | D | Week | 80 | May, 04 2015 00:00:00 | 1 | 80 |
| 1 | 1 | I | D | Week | 80 | May, 05 2015 00:00:00 | 2 | 80 |
| 1 | 1 | I | D | Week | 80 | May, 06 2015 00:00:00 | 3 | 80 |
| 1 | 1 | I | D | Week | 80 | May, 07 2015 00:00:00 | 4 | 80 |
| 1 | 1 | I | D | Week | 80 | May, 08 2015 00:00:00 | 5 | 80 |
| 1 | 1 | I | D | Week | 80 | May, 09 2015 00:00:00 | 6 | 80 |
| 1 | 1 | I | D | Week | 80 | May, 10 2015 00:00:00 | 7 | 80 |
| 1 | 1 | Q | C | Week | 200 | May, 04 2015 00:00:00 | 1 | 60 |
| 1 | 1 | Q | C | Week | 200 | May, 05 2015 00:00:00 | 2 | 20 |
| 1 | 1 | Q | C | Week | 200 | May, 06 2015 00:00:00 | 3 | 20 |
| 1 | 1 | Q | C | Week | 200 | May, 07 2015 00:00:00 | 4 | 10 |
| 1 | 1 | Q | C | Week | 200 | May, 08 2015 00:00:00 | 5 | 20 |
| 1 | 1 | Q | C | Week | 200 | May, 09 2015 00:00:00 | 6 | 30 |
| 1 | 1 | Q | C | Week | 200 | May, 10 2015 00:00:00 | 7 | 40 |
| 1 | 1 | S | C | Week | 100 | May, 04 2015 00:00:00 | 1 | 30 |
| 1 | 1 | S | C | Week | 100 | May, 05 2015 00:00:00 | 2 | 10 |
| 1 | 1 | S | C | Week | 100 | May, 06 2015 00:00:00 | 3 | 10 |
| 1 | 1 | S | C | Week | 100 | May, 07 2015 00:00:00 | 4 | 5 |
| 1 | 1 | S | C | Week | 100 | May, 08 2015 00:00:00 | 5 | 10 |
| 1 | 1 | S | C | Week | 100 | May, 09 2015 00:00:00 | 6 | 15 |
| 1 | 1 | S | C | Week | 100 | May, 10 2015 00:00:00 | 7 | 20 |
| 1 | 2 | I | D | Week | 50 | May, 04 2015 00:00:00 | 1 | 50 |
| 1 | 2 | I | D | Week | 50 | May, 05 2015 00:00:00 | 2 | 50 |
| 1 | 2 | I | D | Week | 50 | May, 06 2015 00:00:00 | 3 | 50 |
| 1 | 2 | I | D | Week | 50 | May, 07 2015 00:00:00 | 4 | 50 |
| 1 | 2 | I | D | Week | 50 | May, 08 2015 00:00:00 | 5 | 50 |
| 1 | 2 | I | D | Week | 50 | May, 09 2015 00:00:00 | 6 | 50 |
| 1 | 2 | I | D | Week | 50 | May, 10 2015 00:00:00 | 7 | 50 |
| 1 | 2 | Q | C | Week | 100 | May, 04 2015 00:00:00 | 1 | 30 |
| 1 | 2 | Q | C | Week | 100 | May, 05 2015 00:00:00 | 2 | 10 |
| 1 | 2 | Q | C | Week | 100 | May, 06 2015 00:00:00 | 3 | 10 |
| 1 | 2 | Q | C | Week | 100 | May, 07 2015 00:00:00 | 4 | 5 |
| 1 | 2 | Q | C | Week | 100 | May, 08 2015 00:00:00 | 5 | 10 |
| 1 | 2 | Q | C | Week | 100 | May, 09 2015 00:00:00 | 6 | 15 |
| 1 | 2 | Q | C | Week | 100 | May, 10 2015 00:00:00 | 7 | 20 |
| 1 | 2 | S | C | Week | 150 | May, 04 2015 00:00:00 | 1 | 45 |
| 1 | 2 | S | C | Week | 150 | May, 05 2015 00:00:00 | 2 | 15 |
| 1 | 2 | S | C | Week | 150 | May, 06 2015 00:00:00 | 3 | 15 |
| 1 | 2 | S | C | Week | 150 | May, 07 2015 00:00:00 | 4 | 7.5 |
| 1 | 2 | S | C | Week | 150 | May, 08 2015 00:00:00 | 5 | 15 |
| 1 | 2 | S | C | Week | 150 | May, 09 2015 00:00:00 | 6 | 22.5 |
| 1 | 2 | S | C | Week | 150 | May, 10 2015 00:00:00 | 7 | 30 |
Once you have this, you need to pivot the result, which can be done by simple GROUP BY
Query:
with x as (
select
r_id,
i_id,
metric,
ctype,
timespan,
quantity,
tdate + level - 1 as m_tdate,
level as m_level,
(case ctype
when 'C' then
(case level
when 1 then 0.3
when 2 then 0.1
when 3 then 0.1
when 4 then 0.05
when 5 then 0.1
when 6 then 0.15
when 7 then 0.2
end)
else 1
end) * quantity as m_quantity
from myt
where timespan = 'Week'
connect by level <= 7
and r_id = prior r_id
and i_id = prior i_id
and metric = prior metric
and ctype = prior ctype
and timespan = prior timespan
and prior sys_guid() is not null
UNION ALL
select
r_id,
i_id,
metric,
ctype,
timespan,
quantity,
tdate as m_tdate,
1 as m_level,
quantity as m_quantity
from myt
where timespan = 'Day'
)
select
r_id,
i_id,
m_tdate,
sum(case when metric = 'S' then m_quantity end) S,
sum(case when metric = 'Q' then m_quantity end) Q,
sum(case when metric = 'I' then m_quantity end) I
from x
group by
r_id,
i_id,
m_tdate
order by
r_id,
i_id,
m_tdate
Results:
| R_ID | I_ID | M_TDATE | S | Q | I |
|------|------|-------------------------|--------|--------|-----|
| 1 | 1 | May, 04 2015 00:00:00 | 30 | 60 | 80 |
| 1 | 1 | May, 05 2015 00:00:00 | 10 | 20 | 80 |
| 1 | 1 | May, 06 2015 00:00:00 | 10 | 20 | 80 |
| 1 | 1 | May, 07 2015 00:00:00 | 5 | 10 | 80 |
| 1 | 1 | May, 08 2015 00:00:00 | 10 | 20 | 80 |
| 1 | 1 | May, 09 2015 00:00:00 | 15 | 30 | 80 |
| 1 | 1 | May, 10 2015 00:00:00 | 20 | 40 | 80 |
| 1 | 2 | April, 03 2015 00:00:00 | (null) | (null) | 120 |
| 1 | 2 | May, 04 2015 00:00:00 | 45 | 30 | 50 |
| 1 | 2 | May, 05 2015 00:00:00 | 15 | 10 | 50 |
| 1 | 2 | May, 06 2015 00:00:00 | 15 | 10 | 50 |
| 1 | 2 | May, 07 2015 00:00:00 | 7.5 | 5 | 50 |
| 1 | 2 | May, 08 2015 00:00:00 | 15 | 10 | 50 |
| 1 | 2 | May, 09 2015 00:00:00 | 22.5 | 15 | 50 |
| 1 | 2 | May, 10 2015 00:00:00 | 30 | 20 | 50 |

postgres, add row when a value is missing

Forgive what may be a silly question, but I'm not much of a database guru.
I have a table with three columns. Here's a sample:
stationtest | id_date | val_no3
------------+-------------+---------
27 | 1 |
27 | 2 | 7
27 | 25 |
27 | 50 | 8
27 | 75 | 9
27 | 100 | 10
30 | 1 |
30 | 14 | 7
30 | 25 |
30 | 65 | 8
30 | 75 | 9
30 | 100 | 10
I would like to have a new table that have one row for each value id_date missing and it combines stationtest number,
like this one :
stationtest | id_date | val_no3
------------+-------------+---------
27 | 1 |
27 | 2 | 7
27 | 3 |
27 | 4 |
27 | 5 |
27 | 6 |
27 | (...) |
27 | 25 |
27 | 26 |
27 | 27 |
27 | (...) |
27 | 50 | 8
27 | (...) |
27 | 75 | 9
27 | (...) |
27 | 98 |
27 | 99 |
27 | 100 | 10
30 | 1 |
30 | 2 | 7
30 | 3 |
30 | 4 |
30 | 5 |
30 | 6 |
30 | (...) |
30 | 25 |
30 | 26 |
30 | 27 |
30 | (...) |
30 | 50 | 8
30 | 75 | 9
30 | (...) |
30 | 98 |
30 | 99 |
30 | 100 | 10
I have this query but i don't know how to make it work for each stationtest :
insert into tabletest (id_date)
select i
from generate_series(1, (select max(id_date) from tabletest)) i
left join tabletest on tabletest.id_date = i
where tabletest.id_date is null;
It is possible ? Thank you for help.

Try this:
DO $$
DECLARE
st_test integer;
i integer;
BEGIN
FOR st_test in (SELECT distinct stationtest FROM tabletest) LOOP
EXECUTE 'INSERT INTO test(stationtest, id_date) SELECT $1 as stationtest, generate_series as id_date FROM generate_series((SELECT min(id_date) FROM test), (SELECT max(id_date) FROM test))' USING st_test;
END LOOP;
END;
$$;
I don't have the data handy, but the general format should work.

How can I increment counter when the value in another column changes?

I have the following table
ID
12
12
25
25
78
78
78
And I need to be able to increment the counter value when the ID changes.
ID **COUNTER**
12 1
12 1
25 2
25 2
78 3
78 3
78 3
How can this be done? Is it even possible?

You can use dense_rank():
select id,
dense_rank() over(order by id) Counter
from yourtable
See SQL Fiddle with Demo
Result:
| ID | COUNTER |
----------------
| 12 | 1 |
| 12 | 1 |
| 25 | 2 |
| 25 | 2 |
| 78 | 3 |
| 78 | 3 |
| 78 | 3 |