The aim of this thesis is to apply M/M/1 and M/M/2 queueing system models
to student and working visas applicants. From empirically observed arrival times and
departure times, the arrival and service intensities are estimated. It is obtained that
the estimated service intensities are essentially double for the M/M/1 queueing model
compared to the M/M/2 model. Hence, the estimated trac intensities, as they are
dened in this thesis, are almost the same here, and close to one. Furthermore, the
empirical distribution of the time in the system for the student visas seems to be
surprisingly well approximated by the theoretical ones obtained from the M/M/1 and
M/M/2 models.
The total time in the system for the two models are almost the same, while for
the M/M/2 model the service times are about the double of the service times of the
M/M/1 model which is compensated by shorter waiting times for the M/M/2 queue.
Therefore, there is a struggle in modelling between having a quicker service time with
less servers or a shorter waiting time period with more servers.
Modelling of the total time in the system, which is the important time for the
applicant and is as well the observed one, seems to be indierent to whether it is
M/M/1 modelling or M/M/2 modelling. That indierence is explained by the fact
that the estimated trac intensity is close to one.
A major consideration is that there is a \wall" eect of the data. The \wall" eect
arises from the ratio of the uncompleted and completed changes towards the end of
year 2017. In this thesis the \wall" eect is resolved by removing the last data.
2