Input-Modeling Report
IEE 475- 3:15 p.m. Lab
Names: xxxx xxxxx xxxx xxxxx xxxx xxxxx
Project Description:
Located at the center of Arizona State University’s Tempe campus, the Memorial Union. It is the
one-stop campus hub featuring dozens of dining options, business, banking and meeting rooms.
It serves hundreds of students, staffs, and visitors daily, it is the only service location on campus
occupied with waiting lines left and right. Though, this project’s primarily focus is on the
Memorial Union’s P.O.D. Market.
The Provisions on Demand Market is a full-service convenience store carrying all student
customer needs on campus, from snacks, hot and frozen foods, beverages, classroom supplies
and so forth. It is open seventeen hours Monday through Friday, from 7a.m. to midnight. This
project is entirely focus on the lunchtime peak hours for ASU students, from around 12 p.m. to
1:30 p.m.
The P.O.D. Market’s system that is being observed and studied is the market’s check-out line
during peak hours. It is a system comprised of a single First-In-First-Out waiting line that
averages about 18.8 seconds before the customer is to be seen. The system is then followed up
by five cashier stations which vary in utilization from 7.83% to 28.7%. These stations do not
fully operate, as the employee's alternate between task. For example, five employees could be
checking out customers and five minutes later, only two employees could be checking-out. This
approach from the P.O.D. Market had an average service time of 30.5 seconds. Overall, the
Arizona State University Memorial Union market averages to keep their customers in the system
for 49.3 seconds.
Inputs:
The market’s customer checkout system is generated by ASU students, staff and visitors. This
system is fully reliable on the customers and the employees who are responsible for checking
out. Without the following two primary inputs, there is no system to study. Other detailed inputs
include the number of products being purchased, card or cash customer, and other types of inputs
that involve status inputs such as employee productivity, product barcode condition, and so forth.
Adjustable Decision Variables:
The P.O.D. Market’s checkout system is a single FIFO waiting line followed by 5 cashier
stations that vary in operation. This system could be adjusted in numerous ways, such as:
● Waiting line set-up other than a single FIFO system.
o Base set-up on product being purchased
o Base set-up on payment method
o Have direct waiting lines to cashier stations
o Base set-up on number of products being purchased
● Number of operational cashier stations.
o Keep a consistent number of cashier operators
o Have cashier operators vary depending on length of waiting line
● Accepted payment method.
o Adjust lines according to payment method
o Just accept cash or just card
o Bill ASU account for all students
As shard, a handful of decision variables could be altered to minimize length of waiting queue
and total service time.
Outputs and Potential Performance Measures:
The customers and employees are the primary inputs of this system, they generate interarrival,
service, and waiting times that are outputs. These measurements collected help to calculate
statistical data such averages, variances and standard deviations for the overall existing system
and for each individual cashier stations. Collecting these outputs help to display cashier and
system performance and percentage utilization during peak hours.
State Variable:
This project is addressing the service P.O.D. market offers during peak hours from 12:00 p.m. to
1:30 p.m. This hour and half-time interval is observed to be the time in which the market serves
more than hundred customers within 30 minutes. It is concluded that this time of day is which
most students, staff and visitors plan to have lunch at the Memorial Union.
Research Questions:
1. Is the current number of cashiers redundant?
2. Will hiring younger (on campus students) cashier operators be more efficient and faster
for the system?
3. If we hired on campus students will that results smaller queue or reduce salary costs for
the system since we will be using less cashiers once the service time has been reduced?
4. Will categorizing cashier lines by payment method or number of items purchased reduce
the waiting time?
5. Will having direct lines for each cashier station be more effective by reducing the total
time spent in the system?
Collected Data Description:
● Interarrival Time (s):
The best fit distribution for the collected
interarrival times, was an exponential
distribution. It is this distribution that was
expected, given that exponential distribution
describes the time between events. In this
case, the time between each customer arrival.
Other supportive reasons for this best fit is the
generated shape and as well the p-value
shared in the distribution summary.
The p-value in the distribution summary on
the right for the Chi-Square Test indicates that
there is no significant evidence to reject the
null hypothesis. In this case, the null
hypothesis is that the corresponding collected
data is of this distribution.
● Service Time (s):
A Gamma Distribution seems reasonable to be
the best-fit distribution for this set of Service
Time data. Yes, it does have a large p-value,
0.467, for the Chi-Square Test and has a
promising shape for this distribution. Though
those aren’t the only supportive details for this
dataset.
Knowing gamma distribution arises naturally
in processes for which the waiting times
between Poisson distributed events are
relevant. Helps to support that this distribution
is the best-fit, given that this data is similar to
the following description of gamma
distribution.
● Total System Time (s):
A Weibull distribution was determined to be
the best-fit distribution for the Total System
Time data collected. It does fit the description
that this distribution is measuring the system’s
reliability, and that is exactly what this data
collected displays.
Another thing to conclude is p-value, which is
again indicating there is no significant
evidence to prove that this distribution in no
the correct distribution. Other than gained p-
values from the Chi-Square and K-S test,
there is the value of the square error.
The square error, measures the differences
between values predicted by a model and the
actual observed. So yes, a small square error
indicates a positive selected best-fit
distribution.
● Waiting Time (s):
As explained for the Service Time dataset,
this set of dataset pertaining to Waiting Time
has a gamma distribution for the same number
of reasons. It meets the description of what is
gamma distribution, it has a corresponding
correct shape and of course has a very
acceptable p-value.
A p-value larger than 0.05 indicates that there
is no significant data to support that the
gamma distribution is the incorrect
distribution. Another reason to accept this
distribution is the value of the Square Error,
it’s small value which is a good sign
according the description provided in the
Total System Time section.