Skip to main content
Announcements
Join us at Qlik Connect for 3 magical days of learning, networking,and inspiration! REGISTER TODAY and save!
cancel
Showing results for 
Search instead for 
Did you mean: 
Sharbel
Contributor III
Contributor III

Join two tables in order to calculate working hours

Hello,

I am trying to import working hours data from csv file  as shown below :

Sharbel_2-1682575475666.png

column a - employee number

column b - date

column c - time 

column d - 100 stands for IN and 200 stands for OUT

IN order to calculate working hours per employee, i thought of creating two tables as shown below:

Table containing only IN's:

Sharbel_4-1682577972497.png

and second table containing the OUT's:

Sharbel_5-1682578052286.png

Note that beside morning and evening shifts (that starts and end within the same day) , there are unique types of working shifts as follows:

Double Shift -  as shown below (in yellow) first shift started at between 07:00 - 10:00 and second shift between 14:00 - 18:00 (start and end may change) 

Sharbel_9-1682578476096.png

Sharbel_10-1682578517769.png

 

Night Shift - usually starts at between 22:00-23:00 and ends at 06:00-07:00 the next day.

Sharbel_7-1682578320607.png

Sharbel_8-1682578357952.png

Any ideas on how to calculate working hours?

 

Many thanks

Sharbel

 

 

 

 

 

 

 

 

 

 

 

Labels (1)
2 Solutions

Accepted Solutions
edwin
Master II
Master II

here is one solution.  i separated the source tables and the transformed table so you can see them at the same time but you can optimize this for sure.  you need to transform your times into the acceptable format - that should be easy enough.  you then inner join them by employee 
then filter out the records where time out < time in
then group by emp and timein and get the min of the timeout.

empIN:
load *, date(date(date_in) + Timestamp(time_in, 'M/D/YYYY hh:mm:ss[.fff]'),'M/D/YYYY hh:mm:ss[.fff]') as date_time_in inline [
emp_in, date_in, time_in
1995, 1/5/2023, 19:08:02
1995, 1/4/2023, 09:08:02
];

empOUT:
load *, date(date(date_out) + Timestamp(time_out, 'M/D/YYYY hh:mm:ss[.fff]'),'M/D/YYYY hh:mm:ss[.fff]') as date_time_out inline [
emp_out, date_out, time_out
1995, 1/6/2023, 07:08:02
1995, 1/4/2023, 23:08:02
];

NoConcatenate
tmp:
load 
	emp_in as emp, date_time_in as datetimein
Resident
	empIN;

inner join (tmp)
load 
	emp_out as emp, date_time_out as datetimeout
resident
	empOUT;
NoConcatenate
EMP:
load 
	emp, datetimein,
    date(min(datetimeout),'M/D/YYYY hh:mm:ss[.fff]') as datetimeout
Resident
	tmp
where datetimeout>=datetimein
group by emp, datetimein
;

drop table tmp;

here is how it looks:

edwin_0-1682619456807.png

result is in hours, you can always floor it after deciding if you will show your duration in hrs, mins, etc

 

View solution in original post

edwin
Master II
Master II

i should have added that the actual in and out should be date time format so the load does that first.  this way it address concerns where in and out are not on the same DAY

View solution in original post

3 Replies
edwin
Master II
Master II

here is one solution.  i separated the source tables and the transformed table so you can see them at the same time but you can optimize this for sure.  you need to transform your times into the acceptable format - that should be easy enough.  you then inner join them by employee 
then filter out the records where time out < time in
then group by emp and timein and get the min of the timeout.

empIN:
load *, date(date(date_in) + Timestamp(time_in, 'M/D/YYYY hh:mm:ss[.fff]'),'M/D/YYYY hh:mm:ss[.fff]') as date_time_in inline [
emp_in, date_in, time_in
1995, 1/5/2023, 19:08:02
1995, 1/4/2023, 09:08:02
];

empOUT:
load *, date(date(date_out) + Timestamp(time_out, 'M/D/YYYY hh:mm:ss[.fff]'),'M/D/YYYY hh:mm:ss[.fff]') as date_time_out inline [
emp_out, date_out, time_out
1995, 1/6/2023, 07:08:02
1995, 1/4/2023, 23:08:02
];

NoConcatenate
tmp:
load 
	emp_in as emp, date_time_in as datetimein
Resident
	empIN;

inner join (tmp)
load 
	emp_out as emp, date_time_out as datetimeout
resident
	empOUT;
NoConcatenate
EMP:
load 
	emp, datetimein,
    date(min(datetimeout),'M/D/YYYY hh:mm:ss[.fff]') as datetimeout
Resident
	tmp
where datetimeout>=datetimein
group by emp, datetimein
;

drop table tmp;

here is how it looks:

edwin_0-1682619456807.png

result is in hours, you can always floor it after deciding if you will show your duration in hrs, mins, etc

 

edwin
Master II
Master II

i should have added that the actual in and out should be date time format so the load does that first.  this way it address concerns where in and out are not on the same DAY

Sharbel
Contributor III
Contributor III
Author

Works Perfect ! 

Thanks a lot