C h a p t e r 4
ANALYSIS OF CHECKOUT SALES OPERATION SERVICE IN ICA
A sales checkout service has 5 waiting lines in a form of parallel cash counters (see fig.1 in the chapter 3).
Customers are served on a first-come, first-served (FIFO) basis as a salesman of checkout operation unit
becomes free. The data has been collected for only two out of five servers on Wednesday (weekday) by
using questionnaires (Appendix A). It was assumed that the customers’ crowd is more, on average, on
weekday. Although the sales checkout unit has 5 parallel counters out of which 2 were observed (each of
them has an individual salesman to deal with the customers in a queue), it is possible that some of the
checkout units are idle. The data collected from questionnaires were tabulated in a spreadsheet in order to
calculate the required parameters of queuing theory analysis (Appendix B). Firstly, the confidence
intervals are computed to estimate service rate and arrival rate for the customers. Then the later first part
of the analysis is done for the model involving one queue and 2 parallel servers (fig.1), whereas the second
part is done by queuing simulation for second model involving 2 queues for each corresponding parallel
server (fig.2).
We can estimate confidence intervals for average service rate and average arrival rate. Assuming service
time and arrival time are iid with N(0,1), then the 95% confidence interval for arrival rate can be:
Confidence Intervals
[( ) ( ) ]
1 1
96.1 ( ) , 96.1 ( )
− − mean arrival time + ⋅ SE mean arrival time mean arrival time − ⋅ SE mean arrival time
where SE(mean arrival time) = SD(mean arrival time /) n
Similarly, 95% confidence interval for service rate can be:
[( ) ( ) ]
1 1
96.1 ( ) , 96.1 ( )
− − mean service time + ⋅ SE mean service time mean service time − ⋅ SE mean service time
where SE(mean servcie time) = SD(mean servcie time /) n