Imagine a sequence of actions. Each action is of a certain type.
Grouping by a given time-frame (e.g. day), how many of these actions happened for the first time, and how many were recurring?
Example Input:
+-----------+-------------+-------------------+
| user_id | action_type | timestamp |
+-----------+-------------+-------------------+
| 5 | play | 2014-02-02 00:55 |
| 2 | sleep | 2014-02-02 00:52 |
| 5 | play | 2014-02-02 00:42 |
| 5 | eat | 2014-02-02 00:31 |
| 3 | eat | 2014-02-02 00:19 |
| 2 | eat | 2014-02-01 23:52 |
| 3 | play | 2014-02-01 23:50 |
| 2 | play | 2014-02-01 23:48 |
+-----------+-------------+-------------------+
Example Output
+------------+------------+-------------+
| first_time | recurring | day |
+------------+------------+-------------+
| 4 | 1 | 2014-02-02 |
| 3 | 0 | 2014-02-01 |
+------------+------------+-------------+
Explanation
On 2014-02-02, users 2, 3, and 5 performed various different actions. There were 4 instances were the users performed an action for the first time; in one case the user 5 repeated the action 'play'.
I added a column 'Total Actions' because as I said, I believe there is a misinterpretation of facts in output example. You can remove it easily.
TEST in SQLFiddle.com for SQL Server 2008.
select
COUNT(q.repetitions) 'first time',
SUM(case when q.repetitions>1 then q.repetitions-1 else 0 end) as 'recurring',
day
from (
select COUNT(i.action_type) as 'repetitions',convert(date,i.time_stamp) as 'day'
from input i
group by i.user_id, i.action_type,convert(date,i.time_stamp)
) q
group by q.day
order by day desc
Related
I have three tables.
The first one is Device table
+----------+------+
| DeviceId | Type |
+----------+------+
| 1 | 10 |
| 2 | 20 |
| 3 | 30 |
+----------+------+
The second one is History table - data received by different devices.
+----------+-------------+--------------------+
| DeviceId | Temperature | TimeStamp |
+----------+-------------+--------------------+
| 1 | 31 | 15.08.2020 1:42:00 |
| 2 | 100 | 15.08.2020 1:42:01 |
| 2 | 40 | 15.08.2020 1:43:00 |
| 1 | 32 | 15.08.2020 1:44:00 |
| 1 | 34 | 15.08.2020 1:45:00 |
| 3 | 20 | 15.08.2020 1:46:00 |
| 2 | 45 | 15.08.2020 1:47:00 |
+----------+-------------+--------------------+
The third one is DeviceStatusHistory table
+----------+---------+--------------------+
| DeviceId | State | TimeStamp |
+----------+---------+--------------------+
| 1 | 1(OK) | 15.08.2020 1:42:00 |
| 2 | 1(OK) | 15.08.2020 1:43:00 |
| 1 | 1(OK) | 15.08.2020 1:44:00 |
| 1 | 0(FAIL) | 15.08.2020 1:44:30 |
| 1 | 0(FAIL) | 15.08.2020 1:46:00 |
| 2 | 0(FAIL) | 15.08.2020 1:46:10 |
+----------+---------+--------------------+
I want to select the last temperature of devices, but take into account only those history records that occurs until the first device failure.
Since device1 starts failing from 15.08.2020 1:44:30, I don't want its records that go after that timestamp.
The same for the device2.
So as a final result, I want to have only data of all devices until they get first FAIL status:
+----------+-------------+--------------------+
| DeviceId | Temperature | TimeStamp |
+----------+-------------+--------------------+
| 2 | 40 | 15.08.2020 1:43:00 |
| 1 | 32 | 15.08.2020 1:44:00 |
| 3 | 20 | 15.08.2020 1:46:00 |
+----------+-------------+--------------------+
I can select an appropriate history only if device failed at least once:
SELECT * FROM Device D
CROSS APPLY
(SELECT TOP 1 * FROM History H
WHERE D.Id = H.DeviceId
and H.DeviceTimeStamp <
(select MIN(UpdatedOn) from DeviceStatusHistory Y where [State]=0 and DeviceId=D.Id)
ORDER BY H.DeviceTimeStamp desc) X
ORDER BY D.Id;
The problems is, if a device never fails, I don't get its history at all.
Update:
My idea is to use something like this
SELECT * FROM DeviceHardwarePart HP
CROSS APPLY
(SELECT TOP 1 * FROM History H
WHERE HP.Id = H.DeviceId
and H.DeviceTimeStamp <
(select ISNULL((select MIN(UpdatedOn) from DeviceMetadataPart where [State]=0 and DeviceId=HP.Id),
cast('12/31/9999 23:59:59.997' as datetime)))
ORDER BY H.DeviceTimeStamp desc) X
ORDER BY HP.Id;
I'm not sure whether it is a good solution
You can use COALESCE: coalesce(min(UpdateOn), cast('9999-12-31 23:59:59' as datetime)). This ensures you always have an upperbound for your select instead of NULL.
I will treat this as two parts problem
I will try to find the time at which device has failed and if it hasn't failed I will keep it as a large value like some timestamp in 2099
Once I have the above I can simply join with histories table and take the latest value before the failed timestamp.
In order to get one, I guess there can be several approaches. From top of my mind something like below should work
select device_id, coalesce(min(failed_timestamps), cast('01-01-2099 01:01:01' as timestamp)) as failed_at
(select device_id, case when state = 0 then timestamp else null end as failed_timestamps from History) as X
group by device_id
This gives us the minimum of failed timestamp for a particular device, and an arbitrary large value for the devices which have never failed.
I guess after this the solution is straight forward.
I have a table that has the data about user_ids, all their last log_in dates to the app
Table:
|----------|--------------|
| User_Id | log_in_dates |
|----------|--------------|
| 1 | 2021-09-01 |
| 1 | 2021-09-03 |
| 2 | 2021-09-02 |
| 2 | 2021-09-04 |
| 3 | 2021-09-01 |
| 3 | 2021-09-02 |
| 3 | 2021-09-03 |
| 3 | 2021-09-04 |
| 4 | 2021-09-03 |
| 4 | 2021-09-04 |
| 5 | 2021-09-01 |
| 6 | 2021-09-01 |
| 6 | 2021-09-09 |
|----------|--------------|
From the above table, I'm trying to understand the user's log in behavior from the present day to the past 90 days.
Num_users_no_log_in defines the count for the number of users who haven't logged in to the app from present_day to the previous days (last_log_in_date)
I want the table like below:
|---------------|------------------|--------------------|-------------------------|
| present_date | days_difference | last_log_in_date | Num_users_no_log_in |
|---------------|------------------|--------------------|-------------------------|
| 2021-09-01 | 0 | 2021-09-01 | 0 |
| 2021-09-02 | 1 | 2021-09-01 | 3 |->(Id = 1,5,6)
| 2021-09-02 | 0 | 2021-09-02 | 3 |->(Id = 1,5,6)
| 2021-09-03 | 2 | 2021-09-01 | 2 |->(Id = 5,6)
| 2021-09-03 | 1 | 2021-09-02 | 1 |->(Id = 2)
| 2021-09-03 | 0 | 2021-09-03 | 3 |->(Id = 2,5,6)
| 2021-09-04 | 3 | 2021-09-01 | 2 |->(Id = 5,6)
| 2021-09-04 | 2 | 2021-09-02 | 0 |
| 2021-09-04 | 1 | 2021-09-03 | 1 |->(Id= 1)
| 2021-09-04 | 0 | 2021-09-04 | 3 |->(Id = 1,5,6)
| .... | .... | .... | ....
|---------------|------------------|--------------------|-------------------------|
I was able to get the first three columns Present_date | days_difference | last_log_in_date using the following query:
with dts as
(
select distinct log_in from users_table
)
select x.log_in_dates as present_date,
DATEDIFF(DAY, y.log_in_dates ,x.log_in_dates ) as Days_since_last_log_in,
y.log_in_dates as log_in_dates
from dts x, dts y
where x.log_in_dates >= y.log_in_dates
I don't understand how I can get the fourth column Num_users_no_log_in
I do not really understand your need: are there values base on users or dates? It it's based on dates, as it looks like (elsewhere you would probably have user_id as first column), what does it mean to have multiple times the same date? I understand that you would like to have a recap for all dates since the beginning until the current date, but in my opinion in does not really make sens (imagine your dashboard in 1 year!!)
Once this is said, let's go to the approach.
In such cases, I develop step by step using common table extensions. For you example, it required 3 steps:
prepare the time series
integrate connections' dates and perform the first calculation (time difference)
Finally, calculate nb connection per day
Then, the final query will display the desired result.
Here is the query I proposed, developed with Postgresql (you did not precise your dbms, but converting should not be such a big deal here):
with init_calendar as (
-- Prepare date series and count total users
select generate_series(min(log_in_dates), now(), interval '1 day') as present_date,
count(distinct user_id) as nb_users
from users
),
calendar as (
-- Add connections' dates for each period from the beginning to current date in calendar
-- and calculate nb days difference for each of them
-- Syntax my vary depending dbms used
select distinct present_date, log_in_dates as last_date,
extract(day from present_date - log_in_dates) as days_difference,
nb_users
from init_calendar
join users on log_in_dates <= present_date
),
usr_con as (
-- Identify last user connection's dates according to running date
-- Tag the line to be counted as no connection
select c.present_date, c.last_date, c.days_difference, c.nb_users,
u.user_id, max(log_in_dates) as last_con,
case when max(log_in_dates) = present_date then 0 else 1 end as to_count
from calendar c
join users u on u.log_in_dates <= c.last_date
group by c.present_date, c.last_date, c.days_difference, c.nb_users, u.user_id
)
select present_date, last_date, days_difference,
nb_users - sum(to_count) as Num_users_no_log_in
from usr_con
group by present_date, last_date, days_difference, nb_users
order by present_date, last_date
Please note that there is a difference with your own expected result as you forgot user_id = 3 in your calculation.
If you want to play with the query, you can with dbfiddle
I'm trying to perform a query between two different tables and come up with a case by case scenario, coming up with a list of records of calls for a specific month.
Here are my tables:
Customer table:
+----+----------------+------------+
| id | name | number |
+----+----------------+------------+
| 1 | John Doe | 8973221232 |
| 2 | American Dad | 7165531212 |
| 3 | Michael Clean | 8884731234 |
| 4 | Samuel Gatsby | 9197543321 |
| 5 | Mike Chat | 8794029819 |
+----+----------------+------------+
Transaction data:
+----------+------------+------------+----------+---------------------+
| trans_id | incoming | outgoing | duration | date_time |
+----------+------------+------------+----------+---------------------+
| 1 | 8973221232 | 9197543321 | 64 | 2018-03-09 01:08:09 |
| 2 | 3729920490 | 7651113929 | 276 | 2018-07-20 05:53:10 |
| 3 | 8884731234 | 8973221232 | 382 | 2018-05-02 13:12:13 |
| 4 | 8973221232 | 9234759208 | 127 | 2018-07-07 15:32:30 |
| 5 | 7165531212 | 9197543321 | 852 | 2018-08-02 07:40:23 |
| 6 | 8884731234 | 9833823023 | 774 | 2018-07-03 14:27:52 |
| 7 | 8273820928 | 2374987349 | 120 | 2018-07-06 05:27:44 |
| 8 | 8973221232 | 9197543321 | 79 | 2018-07-30 12:51:55 |
| 9 | 7165531212 | 7651113929 | 392 | 2018-05-22 02:27:38 |
| 10 | 5423541524 | 7165531212 | 100 | 2018-07-21 22:12:20 |
| 11 | 9197543321 | 2983479820 | 377 | 2018-07-20 17:46:36 |
| 12 | 8973221232 | 7651113929 | 234 | 2018-07-09 03:32:53 |
| 13 | 7165531212 | 2309483932 | 88 | 2018-07-16 16:22:21 |
| 14 | 8973221232 | 8884731234 | 90 | 2018-09-03 13:10:00 |
| 15 | 3820838290 | 2093482348 | 238 | 2018-04-12 21:59:01 |
+----------+------------+------------+----------+---------------------+
What am I trying to accomplish?
I'm trying to compile a list of "costs" for each of the customers that made calls on July 2018. The costs are based on:
1) If the customer received a call (incoming), the cost of the call is equal to the duration;
2) if the customer made a call (outgoing), the cost of the call is 100 if the call is 30 or less in duration. If it exceeds 30 duration, then the cost is 100 plus 5 * duration of the exceeded period.
If the customer didn't make any calls during that month he shouldn't be on the list.
Examples:
1) Customer American Dad has 3 incoming calls and 1 outgoing call, however only trans_id 10 and 13 are for the month of July. He should be paying a total of 538:
for trans_id 10 = 450 (100 for the first 30s + 5 * 70 for the remaining)
for trans_id 13 = 88
2) Customer Samuel Gatsby has 1 incoming call and 3 outgoing calls, however only trans_id 8 and 11 are for the month of July. He should be paying a total of 722:
for trans_id 8 = 345 (100 for the first 30s + 5 * 49 for the remaining)
for trans_id 11 = 377
Considering only these two examples, the output would be:
+----+----------------+------------+------------+
| id | name | number | billable |
+----+----------------+------------+------------+
| 2 | American Dad | 7165531212 | 538 |
| 4 | Samuel Gatsby | 9197543321 | 722 |
+----+----------------+------------+------------+
Note: Mike Chat shouldn't be on the list as he didn't make or receive any calls for that specific month.
What have I tried so far?
I've been playing cat and mouse with this one, I'm using the number as uniqueID, already attempted both a full outer join and combining where incoming or outgoing is not null then applying rules by case, tried doing a left join and applying cases, but I'm circling around and I can't get to a final list. Whenever I get incoming or outgoing, I'm either not able to apply the case or not able to come with both together. Really appreciate the help!
select customer_name.name, customer_name.number, bill = (CASE
WHEN customer_name.number = transaction_data.incoming then 'sum bill'
else 'multiply and add'
end)
from customer_name
left join transaction_data on customer_name.number = transaction_data.incoming or customer_name.name = transaction_data.outgoing
where strftime('%Y-%m', transaction_data.date_time) = '2018-07'
Note: I'm using sqlite to try it out online but the database is on SQL Server 2012, so I know that I can use a date format much easier, that way, but I'd like to keep as close to T-SQL as possible.
Also tried creating a case to determine whether it's incoming call or outgoing, but I'm only getting incoming as a result, even though trans_id 10 is outgoing:
select name, number, duration, case
when customer_name.number = transaction_data.incoming then 'incoming'
when customer_name.number = transaction_data.outgoing then 'outgoing'
END direction
from customer_name
left join transaction_data on customer_name.number = transaction_data.incoming or customer_name.name = transaction_data.outgoing
where strftime('%Y-%m', transaction_data.date_time) = '2018-07'
Try this:
SELECT
c."name", c.number,
SUM(CASE c.number
WHEN t.incoming THEN t.duration
ELSE IIF(t.duration - 30 < 0, 0, t.duration - 30) * 5 + 100
END) AS billable
FROM Customer AS c INNER JOIN [Transaction] AS t
ON c.number IN(t.incoming, t.outgoing)
WHERE t.date_time >= '20180701' AND t.date_time < '20180801'
GROUP BY c."name", c.number
Output:
| name | number | billable |
+---------------+------------+----------+
| John Doe | 8973221232 | 440 |
| American Dad | 7165531212 | 538 |
| Michael Clean | 8884731234 | 774 |
| Samuel Gatsby | 9197543321 | 722 |
Test it online with SQL Fiddle.
I have two tables from a site similar to SO: one with posts, and one with up/down votes for each post. I would like to select all votes cast on the day that a post was modified.
My tables layout is as seen below:
Posts:
-----------------------------------------------
| post_id | post_author | modification_date |
-----------------------------------------------
| 0 | David | 2012-02-25 05:37:34 |
| 1 | David | 2012-02-20 10:13:24 |
| 2 | Matt | 2012-03-27 09:34:33 |
| 3 | Peter | 2012-04-11 19:56:17 |
| ... | ... | ... |
-----------------------------------------------
Votes (each vote is only counted at the end of the day for anonymity):
-------------------------------------------
| vote_id | post_id | vote_date |
-------------------------------------------
| 0 | 0 | 2012-01-13 00:00:00 |
| 1 | 0 | 2012-02-26 00:00:00 |
| 2 | 0 | 2012-02-26 00:00:00 |
| 3 | 0 | 2012-04-12 00:00:00 |
| 4 | 1 | 2012-02-21 00:00:00 |
| ... | ... | ... |
-------------------------------------------
What I want to achieve:
-----------------------------------
| post_id | post_author | vote_id |
-----------------------------------
| 0 | David | 1 |
| 0 | David | 2 |
| 1 | David | 4 |
| ... | ... | ... |
-----------------------------------
I have been able to write the following, but it selects all votes on the day before the post modification, not on the same day (so, in this example, an empty table):
SELECT Posts.post_id, Posts.post_author, Votes.vote_id
FROM Posts
LEFT JOIN Votes ON Posts.post_id = Votes.post_id
WHERE CAST(Posts.modification_date AS DATE) = Votes.vote_date;
How can I fix it so the WHERE clause takes the day before Votes.vote_date? Or, if not possible, is there another way?
Depending on which type of database you are using (SQL, Oracle ect..);To take the Previous days votes you can usually just subtract 1 from the date and it will subtract exactly 1 day:
Where Cast(Posts.modification_date - 1 as Date) = Votes.vote_date
or if modification_date is already in date format just:
Where Posts.modification_date - 1 = Votes.vote_date
If you have a site similar to Stack Overflow, then perhaps you also use SQL Server:
SELECT p.post_id, p.post_author, v.vote_id
FROM Posts p LEFT JOIN
Votes v
ON p.post_id = v.post_id
WHERE CAST(DATEDIFF(day, -1, p.modification_date) AS DATE) = v.vote_date;
Different databases have different ways of subtracting one day. If this doesn't work, then your database has something similar.
I found another solution, which is to add a day to Posts.modification_date:
...
WHERE CAST(CEILING(CAST(p.modification_date AS FLOAT)) AS datetime) = v.vote_date
I have the following set of data:
player | score | day
--------+-------+------------
John | 3 | 02-01-2014
John | 5 | 02-02-2014
John | 7 | 02-03-2014
John | 9 | 02-04-2014
John | 11 | 02-05-2014
John | 13 | 02-06-2014
Mark | 2 | 02-01-2014
Mark | 4 | 02-02-2014
Mark | 6 | 02-03-2014
Mark | 8 | 02-04-2014
Mark | 10 | 02-05-2014
Mark | 12 | 02-06-2014
Given two time ranges:
02-01-2014..02-03-2014
02-04-2014..02-06-2014
I need to get average score for each player within a given time range. Ultimate result I'm trying to achieve is this:
player | period_1_score | period_2_score
--------+----------------+----------------
John | 5 | 11
Mark | 4 | 10
The original algorithm I came up with was:
perform SELECT with two values, derived by partitioning the set of scores into two for each time period
over the first SELECT, perform another one, grouping the set by player name.
I'm stuck on step 1: running the following query:
SELECT
player,
AVG(score) OVER (PARTITION BY day BETWEEN '02-01-2014' AND '02-03-2014') AS period_1,
AVG(score) OVER (PARTITION BY day BETWEEN '02-04-2014' AND '02-06-2014') AS period_2;
Gets me incorrect result (note how period1 and period2 average scores scores are the same:
player | period_1_score | period_2_score
--------+----------------+----------------
John | 5 | 5
John | 5 | 5
John | 5 | 5
John | 5 | 5
John | 5 | 5
John | 5 | 5
Mark | 4 | 4
Mark | 4 | 4
Mark | 4 | 4
Mark | 4 | 4
Mark | 4 | 4
Mark | 4 | 4
I think I don't fully understand how window functions work... I have 2 questions:
What is wrong with my query?
How do I do it right?
You don't need window function for this.
Try:
select
player
,avg(case when day BETWEEN '02-01-2014' AND '02-03-2014' then score else null end) as period_1_score
,avg(case when day BETWEEN '02-04-2014' AND '02-06-2014' then score else null end) as period_1_score
from <your data>
group by player