Finding in between time in a list of times - mysql

Question

I have a table that looks like this
userid | eventid | description | date | starttime | endtime
1 1 Event 1 2016-02-02 09:30:00 11:00:00
1 2 Event 2 2016-02-02 13:30:00 15:00:00
1 3 Event 3 2016-02-02 17:30:00 21:00:00
2 4 Event 4 2016-02-03 13:00:00 14:00:00
2 5 Event 5 2016-02-03 15:00:00 16:00:00
I need to find what is the sum of time between the events on the same day by the user.
Like this:
userid | timeBetween
1 05:00:00
2 01:00:00
I should also assume that there may be overlapping times for example event1 starts at 11:00 ends 13:00 and event2 starts 12:00 and ends 14:00 by the same user on the same day. These cases are rare and I believe returning 00:00 here is the appropriate answer.
I solved a similar problem, finding the sum of the length of all events per day.
SELECT *,
SEC_TO_TIME( SUM( TIME_TO_SEC(TIMEDIFF(`endtime`,`starttime`)))) as sumtime
FROM `events`
group by userid, date
order by sumtime desc

Answer 1

Given this sample data:
CREATE TABLE t
(`userid` int, `eventid` int, `description` varchar(7), `date` date, `starttime` time, `endtime` time)
;
INSERT INTO t
(`userid`, `eventid`, `description`, `date`, `starttime`, `endtime`)
VALUES
(1, 1, 'Event 1', '2016-02-02', '09:30:00', '11:00:00'),
(1, 2, 'Event 2', '2016-02-02', '13:30:00', '15:00:00'),
(1, 3, 'Event 3', '2016-02-02', '17:30:00', '21:00:00'),
(2, 4, 'Event 4', '2016-02-03', '13:00:00', '14:00:00'),
(2, 5, 'Event 5', '2016-02-03', '15:00:00', '16:00:00')
;
this query
SELECT userid, SEC_TO_TIME(SUM(TIME_TO_SEC(diff))) AS time_between
FROM (
SELECT
TIMEDIFF(starttime, COALESCE(IF(userid != #prev_userid, NULL, #prev_endtime), starttime)) AS diff,
#prev_endtime := endtime,
#prev_userid := userid AS userid
FROM
t
, (SELECT #prev_endtime := NULL, #prev_userid := NULL) var_init_subquery
ORDER BY userid
) sq
GROUP BY userid;
will return
+--------+--------------+
| userid | time_between |
+--------+--------------+
| 1 | 05:00:00 |
| 2 | 01:00:00 |
+--------+--------------+
Explanation:
In this part
, (SELECT #prev_endtime := NULL, #prev_userid := NULL) var_init_subquery
ORDER BY userid
we initialize our variables. The ORDER BY is very important, since there's no order in a relational database unless you specify it. It is so important, because the SELECT clause processes the rows in this order.
In the SELECT clause the order is also very important. Here
#prev_endtime := endtime,
#prev_userid := userid AS userid
we assign the values of the current row to the variables. Since this happens after this line
TIMEDIFF(starttime, COALESCE(IF(userid != #prev_userid, NULL, #prev_endtime), starttime)) AS diff,
the variables still hold the values of the previous row in the timediff() function. Therefore we also have to use COALESCE(), because in the very first row and when the userid changes, there is no value to calculate the diff from. To get a diff of 0 there, COALESCE() exchanges the NULL value with the starttime.
The last part is obviously to simply sum the seconds of the "between times".

Answer 2

Here's one way you can get the timeBetween value in SECONDS
SELECT
firsttable.userid,
SEC_TO_TIME(SUM(TIME_TO_SEC(secondtable.starttime) - TIME_TO_SEC(firsttable.endtime))) timeBetween
FROM
(
SELECT
*,
IF(#prev = userid, #rn1 := #rn1 + 1, #rn1 := 1) rank,
#prev := userid
FROM eventtable,(SELECT #prev := 0,#rn1 := 1) var
ORDER BY userid,starttime DESC
) firsttable
INNER JOIN
(
SELECT
*,
IF(#prev2 = userid, #rn2 := #rn2 + 1, #rn2 := 1) rank,
#prev2 := userid
FROM eventtable,(SELECT #prev2 := 0,#rn2 := 1) var
ORDER BY userid,endtime DESC
) secondTable
ON firsttable.userid = secondtable.userid AND firsttable.rank = secondtable.rank + 1 AND
firsttable.date = secondtable.date
GROUP BY firsttable.userid;
TEST:
Unable to add a fiddle.
So here's test data with schema:
DROP TABLE IF EXISTS `eventtable`;
CREATE TABLE `eventtable` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`userid` int(11) NOT NULL,
`eventid` int(11) NOT NULL,
`description` varchar(100) CHARACTER SET utf8 NOT NULL,
`date` date NOT NULL,
`starttime` time NOT NULL,
`endtime` time NOT NULL,
PRIMARY KEY (`id`)
) ;
INSERT INTO `eventtable` VALUES ('1', '1', '1', 'Event 1', '2016-02-02', '09:30:00', '11:00:00');
INSERT INTO `eventtable` VALUES ('2', '1', '2', 'Event 2', '2016-02-02', '13:30:00', '15:00:00');
INSERT INTO `eventtable` VALUES ('3', '1', '3', 'Event 3', '2016-02-02', '17:30:00', '21:00:00');
INSERT INTO `eventtable` VALUES ('4', '2', '4', 'Event 4', '2016-02-03', '13:00:00', '14:00:00');
INSERT INTO `eventtable` VALUES ('5', '2', '5', 'Event 5', '2016-02-03', '15:00:00', '16:00:00');
Result:
Executing the above query on the given test data you will get output like below:
userid timeBetween
1 05:00:00
2 01:00:00
Note:
For overlapping events the above query will give you negative timeBetween value.
You can replace the the SEC_TO_TIME...line by the following:
SEC_TO_TIME(IF(SUM(TIME_TO_SEC(secondtable.starttime) - TIME_TO_SEC(firsttable.endtime)) < 0, 0,SUM(TIME_TO_SEC(secondtable.starttime) - TIME_TO_SEC(firsttable.endtime)))) timeBetween

Answer 3

If you take the TIMEDIFF of the MIN(starttime) and MAX(endtime) for each user/day and then subtract the sum of events as calculated earlier, this will give you the times in between.

Answer 4

try this on
select TIMEDIFF('start_time','end_time') from your table
hope this one help you