I have a table of events, called tbl_events that looks something like this:
PersonID Date
1 30/03/2015
1 22/04/2015
1 30/06/2015
2 18/07/2016
2 09/12/2016
2 28/04/2017
3 01/10/2014
3 28/11/2016
3 28/11/2016
3 16/01/2017
4 13/04/2017
4 09/05/2017
I want to be able to group these events up by the start date of each 'sequence', with a sequence being defined as a run of events from the first identified to the last identified for each PersonID. The last event in a sequence is defined as the event where thereafter there are no subsequent events for that PersonID for a year.
The result of this I would expect to look like is below:
PersonID FirstDate Sequence Events
1 30/03/2015 1 3
2 18/07/2016 1 3
3 01/10/2014 1 1
3 28/11/2016 2 3
4 13/04/2017 1 2
I am able to identify the sequences in Excel and pivot the data, but I need to be able to do this in SQL.
Here is the formula I have used in Excel to generate the sequence number (I am populating cell C3, with column A being PersonID and B being Date):
=+IF(A2<>A3,1,IF((B3-B2)<365,C2,C2+1))
I have joined the table back on itself using ROW_NUMBER to get the difference between the Date and the previous event date for that ID, but I'm not really sure where to go from there.
Any help is much appreciated.
My solution is based on the sample data you've provided along with your excel formula.
-- easily consumable sample data
DECLARE #tbl_events TABLE (PersonId int, [date] date)
INSERT #tbl_events VALUES
(1,'20150330'),(1,'20150422'),(1,'20150630'),(2,'20160718'),(2,'20161209'),(2,'20170428'),
(3,'20141001'),(3,'20161128'),(3,'20161128'),(3,'20170116'),(4,'20170413'),(4,'20170509');
-- Solution
WITH groupings AS
(
SELECT
PersonId,
FirstDate = MIN([date]) OVER (PARTITION BY personId ORDER BY [date]),
NextDate = LAG([date],1,[date]) OVER (PARTITION BY personId ORDER BY [date]),
[date],
grouper =
DATEDIFF(DAY, MIN([date]) OVER (PARTITION BY personId ORDER BY [date]), [date]) / 365
FROM #tbl_events
),
Prep AS
(
SELECT
PersonId,
firstDate = IIF(grouper = 0, FirstDate, IIF(FirstDate = NextDate, [date],NextDate))
FROM groupings
)
SELECT
PersonId,
FirstDate,
[Sequence] = ROW_NUMBER() OVER (PARTITION BY personId ORDER BY FirstDate),
[Events] = COUNT(*)
FROM prep
GROUP BY personId, FirstDate;
Results
PersonId FirstDate Sequence Events
----------- ---------- -------------------- -----------
1 2015-03-30 1 3
2 2016-07-18 1 3
3 2014-10-01 1 1
3 2016-11-28 2 3
4 2017-04-13 1 2
First note all years have 365 days, nonetheless, I'm using 365 to emulate your excel logic; this would need to be updated to account for leap years. Next, like your excel formula - this will only be correct when there are two sequences;
it would not work when, say personId has a date of jan 1 2015, then jan 10 2016, then feb 1 2017.Let us know if we need logic to accommodate for the aforementioned scenarios.
Lastly this solution uses LAG which requires SQL Server 2012+, if you're working with an earlier version of SQL the query will have to be updated accordingly.
Related
I am trying to get a list of:
all months in a specified year that,
have at least 2 unique rows based on their date
and ignore specific column values
where I got to is:
SELECT DATE_PART('month', "orderDate") AS month, count(*)
FROM public."Orders"
WHERE "companyId" = 00001 AND "orderNumber" != 1 and DATE_PART('year', ("orderDate")) = '2020' AND "orderNumber" != NULL
GROUP BY month
HAVING COUNT ("orderDate") > 2
The HAVING_COUNT sort of works in place of DISTINCT insofar as I can be reasonably sure that condition filters the condition of data required.
However, being able to use DISTINCT based on a given date within a month would return a more reliable result. Is this possible with Postgres?
A sample line of data from the table:
Sample Input
"2018-12-17 20:32:00+00"
"2019-02-26 14:38:00+00"
"2020-07-26 10:19:00+00"
"2020-10-13 19:15:00+00"
"2020-10-26 16:42:00+00"
"2020-10-26 19:41:00+00"
"2020-11-19 20:21:00+00"
"2020-11-19 21:22:00+00"
"2020-11-23 21:10:00+00"
"2021-01-02 12:51:00+00"
without the HAVING_COUNT this produces
month
count
7
1
10
2
11
3
Month 7 can be discarded easily as only 1 record.
Month 10 is the issue: we have two records. But from the data above, those records are from the same day. Similarly, month 11 only has 2 distinct records by day.
The output should therefore be ideally:
month
count
11
2
We have only two distinct dates from the 2020 data, and they are from month 11 (November)
I think you just want to take the distinct count of dates for each month:
SELECT
DATE_PART('month', orderDate) AS month,
COUNT(DISTINCT orderDate::date) AS count
FROM Orders
WHERE
companyId = 1 AND
orderNumber != 1 AND
DATE_PART('year', orderDate) = '2020'
GROUP BY
DATE_PART('month', orderDate)
HAVING
COUNT(DISTINCT orderDate::date) > 2;
I have a table like this
item_id date number
1 2000-01-01 100
1 2003-03-08 50
1 2004-04-21 10
1 2004-12-11 10
1 2010-03-03 10
2 2000-06-29 1
2 2002-05-22 2
2 2002-07-06 3
2 2008-10-20 4
I'm trying to get the average for each uniq Item_id over the last 3 dates.
It's difficult because there are missing date in between so a range of hardcoded dates doesn't always work.
I expect a result like :
item_id MyAverage
1 10
2 3
I don't really know how to do this. Currently i manage to do it for one item but i have trouble extending it to multiples items :
SELECT AVG(MyAverage.number) FROM (
SELECT date,number
FROM item_list
where item_id = 1
ORDER BY date DESC limit 3
) as MyAverage;
My main problem is with generalising the "DESC limit 3" over a group by id.
attempt :
SELECT item_id,AVG(MyAverage.number)
FROM (
SELECT item_id,date,number
FROM item_list
ORDER BY date DESC limit 3) as MyAverage
GROUP BY item_id;
The limit is messing things up there.
I have made it " work " using between date and date but it's not working as i want because i need a limit and not an hardcoded date..
Can anybody help
You can use row_number() to assign 1 to 3 for the records with the last date for an ID an then filter for that.
SELECT x.item_id,
avg(x.number)
FROM (SELECT il.item_id,
il.number,
row_number() OVER (PARTITION BY il.item_id
ORDER BY il.date DESC) rn
FROM item_list il) x
WHERE x.rn BETWEEN 1 AND 3
GROUP BY x.item_id;
A fair amount of material is available detailing methods utilising dense_rank() and the like to count distinct somethings per month, however, I've been unable to find anything that allows a count of distinct per month which also removes/discounts any id's that have been seen in prior month groups.
The data can be imagined like so:
id (int8 type) | observed time (timestamp utc)
------------------
1 | 2017-01-01
2 | 2017-01-02
1 | 2017-01-02
1 | 2017-02-02
2 | 2017-02-03
3 | 2017-02-04
1 | 2017-03-01
3 | 2017-03-01
4 | 2017-03-01
5 | 2017-03-02
The process of the count can be seen as:
1: in 2017-01 we saw devices 1 and 2 so the count is 2
2: in 2017-02 we saw devices 1, 2 and 3. We know already about devices 1 and 2, but not 3, so the count is 1
3: in 2017-03 we saw devices 1, 3, 4 and 5. We already know about 1 and 3, but not 4 or 5, so the count is 2.
with the desired output being something like:
observed time | count of new id
--------------------------
2017-01 | 2
2017-02 | 1
2017-03 | 2
Explicitly, I am looking to have a new table, with an aggregated month per row, with a count of how many new ids occur within that month that have not been seen at all before.
The IRL case allows devices to be seen more than once in a month, but this shouldn't impact the count. It also uses integer for storage (both positive and negative) of the id, and time periods will be to the second in true timestamps. The size of the data set is also significant.
My initial attempt is along the lines of:
WITH records_months AS (
SELECT *,
date_trunc('month', observed_time) AS month_group
FROM my_table
WHERE observed_time > '2017-01-01')
id_months AS (
SELECT DISTINCT
month_group,
id
FROM records_months
GROUP BY month_group, id)
SELECT *
FROM id-months
However, I'm stuck on the next part i.e counting the number of new ID that were not seen in prior months. I believe the solution might be a window function, but I'm having trouble working out which or how.
First thing I thought of. The idea is to
(innermost query) calculate the earliest month that each id was seen,
(next level up) join that back to the main my_table dataset, and then
(outer query) count distinct ids by month after nulling out the already-seen ids.
I tested it out and got the desired result set. Joining the earliest month back to the original table seemed like the most natural thing to do (vs. a window function). Hopefully this is performant enough for your Redshift!
select observed_month,
-- Null out the id if the observed_month that we're grouping by
-- is NOT the earliest month that the id was seen.
-- Then count distinct id
count(distinct(case when observed_month != earliest_month then null else id end)) as num_new_ids
from (
select t.id,
date_trunc('month', t.observed_time) as observed_month,
earliest.earliest_month
from my_table t
join (
-- What's the earliest month an id was seen?
select id,
date_trunc('month', min(observed_time)) as earliest_month
from my_table
group by 1
) earliest
on t.id = earliest.id
)
group by 1
order by 1;
I want to count number of month for particular user from subscription table. for example user_id = 1 occur 10 times in subscription table like in January it appears 2 time and in February = 0 and again in march = 1 like that
user_id type started_on ended_on
2 P 2009-10-21 2010-03-18
2 F 2010-03-18 2010-03-20
2 P 2010-03-20 2012-05-19
2 F 2012-05-19 till now
This is pretty basic SQL, I reccomend you read some manual before asking it here. About the Aggregate functions for example. But there you go:
If you want one user:
SELECT
count(distinct month)
FROM
subscription
WHERE
user_id=your_user_id_number
If you want every user's:
SELECT
count(distinct month),
user_id
FROM
subscription
GROUP BY
user_id
Edit:
Ok, so you want the month difference between two date columns, here is how you do it with age():
SELECT user_id, extract(YEAR from age(coalesce(ended_on,current_date),started_on)) * 12 + extract(MONTH FROM age(coalesce(ended_on,current_date),started_on))
FROM subscription
I'm trying to query some transactional data to establish the CurrentProductionHours value for each Report at the end of each month.
Providing there has been a transaction for each report in each month, that's pretty straight-forward... I can use something along the lines of the code below to partition transactions by month and then pick out the rows where TransactionByMonth = 1 (effectively, the last transaction for each report each month).
SELECT
ReportId,
TransactionId,
CurrentProductionHours,
ROW_NUMBER() OVER (PARTITION BY [ReportId], [CalendarYear], [MonthOfYear]
ORDER BY TransactionTimestamp desc
) AS TransactionByMonth
FROM
tblSource
The problem that I have is that there will not necessarily be a transaction for every report every month... When that's the case, I need to carry forward the last known CurrentProductionHours value to the month which has no transaction as this indicates that there has been no change. Potentially, this value may need to be carried forward multiple times.
Source Data:
ReportId TransactionTimestamp CurrentProductionHours
1 2014-01-05 13:37:00 14.50
1 2014-01-20 09:15:00 15.00
1 2014-01-21 10:20:00 10.00
2 2014-01-22 09:43:00 22.00
1 2014-02-02 08:50:00 12.00
Target Results:
ReportId Month Year ProductionHours
1 1 2014 10.00
2 1 2014 22.00
1 2 2014 12.00
2 2 2014 22.00
I should also mention that I have a date table available, which can be referenced if required.
** UPDATE 05/03/2014 **
I now have query which is genertating results as shown in the example below but I'm left with islands of data (where a transaction existed in that month) and gaps in between... My question is still similar but in some ways a little more generic - What is the best way to fill gaps between data islands if you have the dataset below as a starting point?
ReportId Month Year ProductionHours
1 1 2014 10.00
1 2 2014 12.00
1 3 2014 NULL
2 1 2014 22.00
2 2 2014 NULL
2 3 2014 NULL
Any advice about how to tackle this would be greatly appreciated!
Try this:
;with a as
(
select dateadd(m, datediff(m, 0, min(TransactionTimestamp))+1,0) minTransactionTimestamp,
max(TransactionTimestamp) maxTransactionTimestamp from tblSource
), b as
(
select minTransactionTimestamp TT, maxTransactionTimestamp
from a
union all
select dateadd(m, 1, TT), maxTransactionTimestamp
from b
where tt < maxTransactionTimestamp
), c as
(
select distinct t.ReportId, b.TT from tblSource t
cross apply b
)
select c.ReportId,
month(dateadd(m, -1, c.TT)) Month,
year(dateadd(m, -1, c.TT)) Year,
x.CurrentProductionHours
from c
cross apply
(select top 1 CurrentProductionHours from tblSource
where TransactionTimestamp < c.TT
and ReportId = c.ReportId
order by TransactionTimestamp desc) x
A similar approach but using a cartesian to obtain all the combinations of report ids/months.
in the first step.
A second step adds to that cartesian the maximum timestamp from the source table where the month is less or equal to the month in the current row.
Finally it joins the source table to the temp table by report id/timestamp to obtain the latest source table row for every report id/month.
;
WITH allcombinations -- Cartesian (reportid X yearmonth)
AS ( SELECT reportid ,
yearmonth
FROM ( SELECT DISTINCT
reportid
FROM tblSource
) a
JOIN ( SELECT DISTINCT
DATEPART(yy, transactionTimestamp)
* 100 + DATEPART(MM,
transactionTimestamp) yearmonth
FROM tblSource
) b ON 1 = 1
),
maxdates --add correlated max timestamp where the month is less or equal to the month in current record
AS ( SELECT a.* ,
( SELECT MAX(transactionTimestamp)
FROM tblSource t
WHERE t.reportid = a.reportid
AND DATEPART(yy, t.transactionTimestamp)
* 100 + DATEPART(MM,
t.transactionTimestamp) <= a.yearmonth
) maxtstamp
FROM allcombinations a
)
-- join previous data to the source table by reportid and timestamp
SELECT distinct m.reportid ,
m.yearmonth ,
t.CurrentProductionHours
FROM maxdates m
JOIN tblSource t ON t.transactionTimestamp = m.maxtstamp and t.reportid=m.reportid
ORDER BY m.reportid ,
m.yearmonth