I have a table with created_at and deleted_at timestamps. I need to know, for each week, how many records existed at any point that week:
week
records
2022-01
4
2022-02
5
...
...
Essentially, records that were created before the end of the week and deleted after the beginning of the week.
I've tried various variations of the following but it's under-reporting and I can't work out why:
SELECT
DATE_FORMAT(created_at, '%Y-%U') AS week,
COUNT(*)
FROM records
WHERE
deleted_at > DATE_SUB(deleted_at, INTERVAL (WEEKDAY(deleted_at)+1) DAY)
AND created_at < DATE_ADD(created_at, INTERVAL 7 - WEEKDAY(created_at) DAY)
GROUP BY week
ORDER BY week
Any help would be massively appreciated!
I would create a table wktable that looks like so (for the last 5 weeks of last year):
yrweek | wkstart | wkstart
-------+------------+------------
202249 | 2022-11-27 | 2022-12-03
202250 | 2022-12-04 | 2022-12-10
202251 | 2022-12-11 | 2022-12-17
202252 | 2022-12-18 | 2022-12-24
202253 | 2022-12-25 | 2022-12-31
To get there, find a way to create 365 consecutive integers, make all the dates of 2022 out of that, and group them by year-week.
This is an example:
CREATE TABLE wk AS
WITH units(units) AS (
SELECT 0 UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION
SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9
)
,tens AS(SELECT units * 10 AS tens FROM units )
,hundreds AS(SELECT tens * 10 AS hundreds FROM tens )
,
i(i) AS (
SELECT hundreds +tens +units
FROM units
CROSS JOIN tens
CROSS JOIN hundreds
)
,
dt(dt) AS (
SELECT
DATE_ADD(DATE '2022-01-01', INTERVAL i DAY)
FROM i
WHERE i < 365
)
SELECT
YEAR(dt)*100 + WEEK(dt) AS yrweek
, MIN(dt) AS wkstart
, MAX(dt) AS wkend
FROM dt
GROUP BY yrweek
ORDER BY yrweek;
With that table, go:
SELECT
yrweek
, COUNT(*) AS records
FROM wk
JOIN input_table ON wk.wkstart < input_table.deleted_at
AND wk.wkend > input_table.created_at
GROUP BY
yrweek
;
I first build a list with the records, their open count, and the closed count
SELECT
created_at,
deleted_at,
(SELECT COUNT(*)
from records r2
where r2.created_at <= r1.created_at ) as new,
(SELECT COUNT(*)
from records r2
where r2.deleted_at <= r1.created_at) as closed
FROM records r1
ORDER BY r1.created_at;
After that it's just adding a GROUP BY:
SELECT
date_format(created_at,'%Y-%U') as week,
MAX((SELECT COUNT(*)
from records r2
where r2.created_at <= r1.created_at )) as new,
MAX((SELECT COUNT(*)
from records r2
where r2.deleted_at <= r1.created_at)) as closed
FROM records r1
GROUP BY week
ORDER BY week;
see: DBFIDDLE
NOTE: Because I use random times, the results will change when re-run. A sample output is:
week
new
closed
2022-00
31
0
2022-01
298
64
2022-02
570
212
2022-03
800
421
Let's say I have the following table:
date | name | value
----------------------------
2020-09-01 | name1 | 10
2020-09-02 | name1 | 9
2020-09-03 | name1 | 12
2020-09-04 | name1 | 11
2020-09-05 | name1 | 11
I would like to identify names where the latest value >= 10 AND where over the last 5 days it has ever dropped below 10. In the example table above, name1 would be returned because the latest date has a value of 11 (which is > 10), and over the last 5 days it has dropped below 10 at least once.
Here is my SELECT statement, but it always returns zero rows:
SELECT
name,
count(value) as count
FROM table_name
WHERE
(date = #date AND value >= 10) AND
date BETWEEN date_sub(#date, interval 5 day) AND #date AND value < 10
GROUP BY name
HAVING count < 5
ORDER BY name
I understand why it's failing, but I don't know what to change.
In MySQL 8.0, you could use window functions and aggregation:
select name
from (
select t.*, row_number() over(partition by name order by date desc) rn
from mytable t
where date >= #date - interval 5 day and date <= #date
) t
having max(case when rn = 1 then value end) >= 10 and min(value) <= 10
How about something like this:
SELECT Name, COUNT(*) AS Ct FROM
(SELECT A.*,B.mdate,
CASE WHEN A.date=B.mdate AND A.value >= 10 THEN 1
WHEN A.date >= B.mdate - INTERVAL 5 DAY AND A.date <> B.mdate AND A.value < 10 THEN 1
ELSE 0 END AS Chk
FROM table_name A
JOIN (SELECT Name,MAX(DATE) AS mdate FROM table_name GROUP BY Name) B ON A.Name=B.Name
HAVING Chk <> 0) V
GROUP BY Name
HAVING Ct >= 2
Here's a fiddle for reference: https://www.db-fiddle.com/f/jX4GktCdTrUbqHBf7ZQwdr/0
And here's a breakdown of what the query above is doing.
Joining table_name with a sub-query of the same table but with MAX(DATE) value for comparison.
Using CASE function to check for your conditions; if matches with the conditions, it will return 1, if not, return 0. Added HAVING to exclude any 0 value from the CASE function.
Turn the query to become a sub-query (assigned as V) and do a COUNT(*) over how many occurrence happen on the name then using HAVING again to get any name that have 2 or more occurrence.
I'm trying to select the count of users retaining in a certain date range and that was successful using this query and the table below:
+----------+-------------+
| Field | Type |
+----------+-------------+
| id | varchar(17) |
| log_date | date |
+----------+-------------+
SELECT last_day.log_date, COUNT(distinct last_day.id) as users_num
FROM (SELECT DISTINCT log_date, id
FROM `userActivity`) this_day
JOIN (SELECT DISTINCT log_date, id
FROM `userActivity`) last_day
ON this_day.id = last_day.id
AND this_day.log_date = "2018-10-01"
AND last_day.log_date BETWEEN "2018-10-01" AND "2018-10-30"
GROUP BY log_date;
But the problem that I'm facing is that I want the assume that every day of the date-range is day 0. (similar to the following example):
Note that the first row in the pic is the avg of the below results I need to calculate. Anyone has any idea how can I enhance my query to get the result like the picture?
This solution will only work on MySQL 8.x only since it requires CTEs (Common Table Expressions):
with digits as (
select 0 as n union select 1 union select 2 union select 3 union select 4
union select 5 union select 6 union select 7 union select 8 union select 9
),
series as (
select d1.n * 100 + d2.n * 10 + d3.n as n -- number series from 0 to 999
from digits d1
cross join digits d2
cross join digits d3
)
SELECT last_day.log_date, COUNT(distinct last_day.id) as users_num,
date_add("2018-10-01", interval s.n day) as current_start
FROM (SELECT DISTINCT log_date, id
FROM `userActivity`) this_day
JOIN (SELECT DISTINCT log_date, id
FROM `userActivity`) last_day
ON this_day.id = last_day.id
cross join series s
WHERE s.n <= 30
AND this_day.log_date = date_add("2018-10-01", interval s.n day)
AND last_day.log_date BETWEEN date_add("2018-10-01", interval s.n day)
AND date_add("2018-10-30", interval s.n day)
GROUP BY log_date, date_add("2018-10-01", interval s.n day);
I am trying to generate a table in the following format.
Proday | 2014-04-01 | 2014-03-01
--------------------------------
1 | 12 | 17
2 | 6 | 0
7 | 0 | 24
13 | 3 | 7
Prodays (duration between two timestamps) is a calculated value and the data for months is a COUNT. I can output the data for a single month, but am having troubles joining queries to additional months. The index (prodays) may not match for each month. e.g.. 2014-04-01 may not have any data for Prodays 7, whereas 2014-03-01 may not have Proday 2. Should indicate with 0 or null.
I suspect FULL OUTER JOIN is what should do the trick. But have read that's not possible in Mysql?
This is the query to get data for a single month:
SELECT round((protime - createtime) / 86400) AS prodays, COUNT(id) AS '2014-04-01'
FROM `tbl_users` as t1
WHERE status = 1 AND DATE_FORMAT(FROM_UNIXTIME(createtime),'%Y-%m-%d') >= '2014-04-01'
AND DATE_FORMAT(FROM_UNIXTIME(createtime),'%Y-%m-%d') <= LAST_DAY('2014-04-01')
GROUP BY prodays
ORDER BY `prodays` ASC
How can I join/union an additional query to create a column for 2014-03-01?
You want to use conditional aggregation -- that is, move the filtering logic from the where clause to the select clause:
SELECT round((protime - createtime) / 86400) AS prodays,
sum(DATE_FORMAT(FROM_UNIXTIME(createtime),'%Y-%m-%d') >= '2014-04-01' AND
DATE_FORMAT(FROM_UNIXTIME(createtime),'%Y-%m-%d') <= LAST_DAY('2014-04-01')
) as `2014-04-01`,
sum(DATE_FORMAT(FROM_UNIXTIME(createtime),'%Y-%m-%d') >= '2014-03-01' AND
DATE_FORMAT(FROM_UNIXTIME(createtime),'%Y-%m-%d') <= LAST_DAY('2014-03-01')
) as `2014-03-01`
FROM `tbl_users` as t1
WHERE status = 1
GROUP BY prodays
ORDER BY `prodays` ASC;
A rental system uses a booking table to store all bookings and reservations:
booking | item | startdate | enddate
1 | 42 | 2013-10-25 16:00 | 2013-10-27 12:00
2 | 42 | 2013-10-27 14:00 | 2013-10-28 18:00
3 | 42 | 2013-10-30 09:00 | 2013-11-01 09:00
…
Let’s say a user wants to rent item 42 from 2013-10-27 12:00 until 2013-10-28 12:00 which is a period of one day. The system will tell him, that the item is not available in the given time frame, since booking no. 2 collides.
Now I want to suggest the earliest rental date and time when the selected item is available again. Of course considering the user’s requested period (1 day) beginning with the user’s desired date and time.
So in the case above, I’m looking for an SQL query that returns 2013-10-28 18:00, since the earliest date since 2013-10-27 12:00 at which item 42 will be available for 1 day, is from 2013-10-28 18:00 until 2013-10-29 18:00.
So I need to to find a gap between bookings, that is big enough to hold the user’s reservation and that is as close a possible to the desired start date.
Or in other words: I need to find the first booking for a given item, after which there’s enough free time to place the user’s booking.
Is this possible in plain SQL without having to iterate over every booking and its successor?
If you can't redesign your database to use something more efficient, this will get the answer. You'll obviously want to parameterize it. It says find either the desired date, or the earliest end date where the hire interval doesn't overlap an existing booking:
Select
min(startdate)
From (
select
cast('2013-10-27 12:00' as datetime) startdate
from
dual
union all
select
enddate
from
booking
where
enddate > cast('2013-10-27 12:00' as datetime) and
item = 42
) b1
Where
not exists (
select
'x'
from
booking b2
where
item = 42 and
b1.startdate < b2.enddate and
b2.startdate < date_add(b1.startdate, interval 24 hour)
);
Example Fiddle
SELECT startfree,secondsfree FROM (
SELECT
#lastenddate AS startfree,
UNIX_TIMESTAMP(startdate)-UNIX_TIMESTAMP(#lastenddate) AS secondsfree,
#lastenddate:=enddate AS ignoreme
FROM
(SELECT startdate,enddate FROM bookings WHERE item=42) AS schedule,
(SELECT #lastenddate:=NOW()) AS init
ORDER BY startdate
) AS baseview
WHERE startfree>='2013-10-27 12:00:00'
AND secondsfree>=86400
ORDER BY startfree
LIMIT 1
;
Some explanation: The inner query uses a variable to move the iteration into SQL, the outer query finds the needed row.
That said, I would not do this in SQL, if the DB structure is like the given. You could reduce the iteration count by using some smort WHERE in the inner query to a sane timespan, but chances are, this won't perform well.
EDIT
A caveat: I did not check, but I assume, this won't work, if there are no prior reservations in the list - this should not be a problem, as in this case your first reservation attempt (original time) will work.
EDIT
SQLfiddle
Searching for overlapping date ranges generally yields poor performance in SQL. For that reason having a "Calendar" of available slots often makes things a lot more efficient.
For example, the booking 2013-10-25 16:00 => 2013-10-27 12:00 would actually be represented by 44 records, each one hour long.
The "gap" until the next booking at 2013-10-27 14:00 would then be represented by 2 records, each one hours long.
Then, each record could also have the duration (in time, or number of slots) until the next change.
slot_start_time | booking | item | remaining_duration
------------------+---------+------+--------------------
2013-10-27 10:00 | 1 | 42 | 2
2013-10-27 11:00 | 1 | 42 | 1
2013-10-27 12:00 | NULL | 42 | 2
2013-10-27 13:00 | NULL | 42 | 1
2013-10-27 14:00 | 2 | 42 | 28
2013-10-27 15:00 | 2 | 42 | 27
... | ... | ... | ...
2013-10-28 17:00 | 2 | 42 | 1
2013-10-28 18:00 | NULL | 42 | 39
2013-10-28 19:00 | NULL | 42 | 38
Then your query just becomes:
SELECT
*
FROM
slots
WHERE
slot_start_time >= '2013-10-27 12:00'
AND remaining_duration >= 24
AND booking IS NULL
ORDER BY
slot_start_time ASC
LIMIT
1
OK this isn't pretty in MySQL. That's because we have to fake rownum values in subqueries.
The basic approach is to join the appropriate subset of the booking table to itself offset by one.
Here's the basic list of reservations for item 42, ordered by reservation time. We can't order by booking_id, because those aren't guaranteed to be in order of reservation time. (You're trying to insert a new reservation between two existing ones, eh?) http://sqlfiddle.com/#!2/62383/9/0
SELECT #aserial := #aserial+1 AS rownum,
booking.*
FROM booking,
(SELECT #aserial:= 0) AS q
WHERE item = 42
ORDER BY startdate, enddate
Here is that subset joined to itself. The trick is the a.rownum+1 = b.rownum, which joins each row to the one that comes right after it in the booking table subset. http://sqlfiddle.com/#!2/62383/8/0
SELECT a.booking_id, a.startdate asta, a.enddate aend,
b.startdate bsta, b.enddate bend
FROM (
SELECT #aserial := #aserial+1 AS rownum,
booking.*
FROM booking,
(SELECT #aserial:= 0) AS q
WHERE item = 42
ORDER BY startdate, enddate
) AS a
JOIN (
SELECT #bserial := #bserial+1 AS rownum,
booking.*
FROM booking,
(SELECT #bserial:= 0) AS q
WHERE item = 42
ORDER BY startdate, enddate
) AS b ON a.rownum+1 = b.rownum
Here it is again, showing each reservation (except the last one) and the number of hours following it. http://sqlfiddle.com/#!2/62383/15/0
SELECT a.booking_id, a.startdate, a.enddate,
TIMESTAMPDIFF(HOUR, a.enddate, b.startdate) gaphours
FROM (
SELECT #aserial := #aserial+1 AS rownum,
booking.*
FROM booking,
(SELECT #aserial:= 0) AS q
WHERE item = 42
ORDER BY startdate, enddate
) AS a
JOIN (
SELECT #bserial := #bserial+1 AS rownum,
booking.*
FROM booking,
(SELECT #bserial:= 0) AS q
WHERE item = 42
ORDER BY startdate, enddate
) AS b ON a.rownum+1 = b.rownum
So, if you're looking for the starting time and ending time of the earliest twelve-hour slot you can use that result set to do this: http://sqlfiddle.com/#!2/62383/18/0
SELECT MIN(enddate) startdate, MIN(enddate) + INTERVAL 12 HOUR as enddate
FROM (
SELECT a.booking_id, a.startdate, a.enddate,
TIMESTAMPDIFF(HOUR, a.enddate, b.startdate) gaphours
FROM (
SELECT #aserial := #aserial+1 AS rownum,
booking.*
FROM booking,
(SELECT #aserial:= 0) AS q
WHERE item = 42
ORDER BY startdate, enddate
) AS a
JOIN (
SELECT #bserial := #bserial+1 AS rownum,
booking.*
FROM booking,
(SELECT #bserial:= 0) AS q
WHERE item = 42
ORDER BY startdate, enddate
) AS b ON a.rownum+1 = b.rownum
) AS gaps
WHERE gaphours >= 12
here is the query, it will return needed date, obvious condition - there should be some bookings in table, but as I see from question - you do this check:
SELECT min(enddate)
FROM
(
select a.enddate from table4 as a
where
a.item=42
and
DATE_ADD(a.enddate, INTERVAL 1 day) <= ifnull(
(select min(b.startdate)
from table4 as b where b.startdate>=a.enddate and a.item=b.item),
a.enddate)
and
a.enddate>=now()
union all
select greatest(ifnull(max(enddate), now()),now()) from table4
) as q
you change change INTERVAL 1 day to INTERVAL ### hour
If I have understood your requirements correctly, you could try self-JOINing book with itself, to get the "empty" spaces, and then fit. This is MySQL only (I believe it can be adapted to others - certainly PostgreSQL):
SELECT book.*, TIMESTAMPDIFF(MINUTE, book.enddate, book.best) AS width FROM
(
SELECT book.*, MIN(book1.startdate) AS best
FROM book
JOIN book AS book1 USING (item)
WHERE item = 42 AND book1.startdate >= book.enddate
GROUP BY book.booking
) AS book HAVING width > 110 ORDER BY startdate LIMIT 1;
In the above example, "110" is the looked-for minimum width in minutes.
Same thing, a bit less readable (for me), a SELECT removed (very fast SELECT, so little advantage):
SELECT book.*, MIN(book1.startdate) AS best
FROM book
JOIN book AS book1 ON (book.item = book1.item AND book.item = 42)
WHERE book1.startdate >= book.enddate
GROUP BY book.booking
HAVING TIMESTAMPDIFF(MINUTE, book.enddate, best) > 110
ORDER BY startdate LIMIT 1;
In your case, one day is 1440 minutes and
SELECT book.*, MIN(book1.startdate) AS best FROM book JOIN book AS book1 ON (book.item = book1.item AND book.item = 42) WHERE book1.startdate >= book.enddate GROUP BY book.booking HAVING TIMESTAMPDIFF(MINUTE, book.enddate, best) >= 1440 ORDER BY startdate LIMIT 1;
+---------+------+---------------------+---------------------+---------------------+
| booking | item | startdate | enddate | best |
+---------+------+---------------------+---------------------+---------------------+
| 2 | 42 | 2013-10-27 14:00:00 | 2013-10-28 18:00:00 | 2013-10-30 09:00:00 |
+---------+------+---------------------+---------------------+---------------------+
1 row in set (0.00 sec)
...the period returned is 2, i.e., at the end of booking 2, and until "best" which is booking 3, a period of at least 1440 minutes is available.
An issue could be that if no periods are available, the query returns nothing -- then you need another query to fetch the farthest enddate. You can do this with an UNION and LIMIT 1 of course, but I think it would be best to only run the 'recovery' query on demand, programmatically (i.e. if empty(query) then new_query...).
Also, in the inner WHERE you should add a check for NOW() to avoid dates in the past. If expired bookings are moved to inactive storage, this could be unnecessary.