Chapter Four
Output Analysis
By: Mr. Getahun M.
Output Analysis
01 02
Terminating & Statistics Collection
Steady-State from Replications
Simulation Models
03 04
Confidence Interval
Point Estimation Estimation
01
Terminating &
Steady-State
Simulation Models
Terminating Simulation Models
● Terminating simulation models are used to
study systems that have a clear beginning and
end.
● For example, a terminating simulation model
could be used to study the performance of a
new manufacturing process over a period of six
months.
Cont’…
● Terminating simulation models are typically used
to answer questions such as:
○ How long will it take to complete the project?
○ What is the probability that the project will be
completed on time?
○ What is the expected cost of the project?
Steady – State Simulation Models
● Steady-state simulation models are used to
study systems that reach a state of equilibrium.
● For example, a steady-state simulation model
could be used to study the performance of a call
center.
Cont’…
● Steady-state simulation models are typically
used to answer questions such as:
○ What is the average wait time for customers?
○ What is the average number of customers in
the system?
○ What is the probability that a customer will
have to wait more than 10 minutes?
Advantage & Disadvantage of Terminating Simulation Models
● Advantage:
○ Terminating simulation models are typically
easier to develop and run than steady-state
simulation models.
○ Terminating simulation models can be used
to study systems that have a clear beginning
and end.
Cont’…
● Disadvantage:
○ Terminating simulation models may not be
able to accurately represent the behaviour of
systems that do not have a clear beginning
and end.
○ Terminating simulation models may not be
able to answer questions about the long-term
behaviour of systems.
Advantage & Disadvantage of Steady – State Simulation Models
● Advantage:
○ Steady-state simulation models can be used
to accurately represent the behaviour of
systems that reach a state of equilibrium.
○ Steady-state simulation models can be used
to answer questions about the long-term
behaviour of systems.
Cont’…
● Disadvantage:
○ Steady-state simulation models are typically
more difficult to develop and run than
terminating simulation models.
○ Steady-state simulation models may not be
able to study systems that have a clear
beginning and end.
02
Statistics Collection
from Replications
Statistics Collection Using Independent
Replications
What are independent replications?
● Independent replications are multiple runs of a
simulation with the same input parameters.
● The output from each replication is independent of
the output from the other replications.
Cont’…
Why use independent replications?
● Independent replications can be used to reduce
the variance of the statistics.
● This can lead to more accurate results.
● Independent replications can also be used to assess
the variability of the simulation model.
Cont’…
How to use independent replications
● There are a few things to keep in mind when
using independent replications:
○ The number of replications should be large
enough to reduce the variance of the statistics to
an acceptable level.
○ The replications should be run independently of
each other.
○ The output from each replication should be saved.
Cont’…
Example
● Consider a simulation of a queue.
● The goal of the simulation is to estimate the
average waiting time in the queue. One way to
collect the average waiting time is to run the
simulation once and collect the waiting time for
each customer who enters the queue. However,
this approach can be computationally expensive,
especially if the queue is long.
Cont’…
● A better approach is to use independent replications.
● In this case, the simulation could be run 100 times.
The average waiting time for each replication could
then be calculated.
● The average waiting time for all replications could
then be calculated by averaging the average waiting
times for each replication.
Statistics Collection Using Regeneration Points &
Batch Means
● In simulation, statistics are often collected over a
long period of time, which can be computationally
expensive. One way to reduce the computational
cost is to use regeneration points.
● A regeneration point is a point in the simulation
where the system is in a steady state. Once a
regeneration point is reached, statistics can be
collected without bias.
Cont’…
● Another way to reduce the computational cost of
collecting statistics is to use batch means.
● Batch means is a technique where statistics are
collected over a group of simulation runs, and
then averaged. This can reduce the variance of
the statistics, which can lead to more accurate
results.
Cont’…
Examples
● In this section, we will provide some examples of
how regeneration points and batch means can
be used in practice.
Cont’…
Example 1
● Consider a simulation of a queue. The goal of the
simulation is to estimate the average waiting time
in the queue.
Cont’…
● One way to collect the average waiting time is to
collect the waiting time for each customer who
enters the queue. However, this approach can be
computationally expensive, especially if the queue
is long.
Cont’…
● A better approach is to use regeneration points. A
regeneration point in this case could be reached
when the queue length reaches zero.
● Once a regeneration point is reached, the average
waiting time for the current batch of customers
can be calculated.
Cont’…
● The average waiting time for all customers can
then be calculated by averaging the average
waiting times for each batch.
Cont’…
Example 2
● Consider a simulation of a manufacturing system.
The goal of the simulation is to estimate the
average production rate of the system.
Cont’…
● One way to collect the average production rate is
to collect the production time for each batch of
products that is produced. However, this approach
can be computationally expensive, especially if the
production time is long.
Cont’…
● A better approach is to use batch means. In this
case, the batch size could be set to the number of
products in a single batch.
● The average production rate for each batch can
then be calculated. The average production rate
for all batches can then be calculated by averaging
the average production rates for each batch.
03
Point Estimation
Cont’…
What is point estimation?
● Point estimation is the process of using
sample data to calculate a single value that is
to serve as a "best guess" or "best estimate" of
an unknown population parameter.
Cont’…
For example,
● If we want to estimate the average waiting
time in a queue, we could run a simulation of
the queue 100 times and calculate the average
waiting time for each run. The average waiting
time for all 100 runs would be our point
estimate of the average waiting time in the
queue.
Cont’…
Types of point estimators
● There are many different types of point
estimators, each with its own advantages and
disadvantages. Some of the most common
types of point estimators include:
Cont’…
○ The sample mean: The sample mean is the
most common type of point estimator. It is
calculated by adding up the values of all the
observations in the sample and dividing by
the number of observations.
Cont’…
○ The sample median: The sample median is the
middle value in a sorted sample. It is a robust
estimator, meaning that it is not as sensitive to
outliers as the sample mean.
Cont’…
○ The sample mode: The sample mode is the value
that appears most often in a sample. It is not as
widely used as the sample mean or the sample
median, but it can be useful for identifying the
most common value in a sample.
Cont’…
Choosing a point estimator
● There is no single "best" point estimator for all
situations. The best point estimator to use will
depend on the specific problem being studied.
Some factors to consider when choosing a point
estimator include:
Cont’…
○ The type of data being collected
○ The level of accuracy desired
○ The presence of outliers
04
Confidence
Interval Estimation
Confidence Intervals for Terminating
Simulations
What is a confidence interval?
● A confidence interval is a range of values that is
likely to contain the true value of a population
parameter.
Cont’…
● For example, if we want to estimate the average
waiting time in a queue, we could calculate a
95% confidence interval for the average waiting
time. This would mean that we are 95% confident
that the true average waiting time in the queue
lies within the confidence interval.
Cont’…
How to calculate a confidence interval for a
terminating simulation.
● There are a few different ways to calculate a
confidence interval for a terminating simulation.
One common approach is to use the central limit
theorem.
Cont’…
● The central limit theorem states that the
distribution of the sample mean will be
approximately normal, even if the distribution of
the population mean is not normal. This means
that we can use the normal distribution to
calculate a confidence interval for the sample
mean.
Cont’…
Example
● Consider a simulation of a queue. The goal of the
simulation is to estimate the average waiting
time in the queue. The simulation is run for 100
iterations. The average waiting time for the 100
iterations is 5 minutes.
Cont’…
● We can use the central limit theorem to
calculate a 95% confidence interval for the
average waiting time. The formula for a 95%
confidence interval is:
o Confidence interval = sample mean ± (1.96
* standard error of the mean)
Cont’…
● The standard error of the mean is calculated as
follows:
o Standard error of the mean = (standard
deviation of the sample) / sqrt(number of
iterations)
Cont’…
● In this case, the standard deviation of the sample
is 1 minute and the number of iterations is 100.
The standard error of the mean is therefore 0.1
minutes.
● Plugging the values of the sample mean,
standard deviation, and number of iterations into
the formula for a 95% confidence interval, we get
the following confidence interval:
Cont’…
o Confidence interval = 5 minutes ± (1.96 * 0.1
minutes)
o confidence interval = 4.04 minutes to 5.96
minutes
● This means that we are 95% confident that the
true average waiting time in the queue lies
between 4.04 minutes and 5.96 minutes.
Confidence Intervals for Steady – State
Simulation
What is steady state?
● Steady state is a state in which the system has
reached a stable equilibrium. This means that
the system's output is no longer changing over
time.
Cont’…
How to calculate a confidence interval for a
steady-state simulation
● There are a few different ways to calculate a
confidence interval for a steady-state simulation.
One common approach is to use the central limit
theorem.
Cont’…
● The central limit theorem states that the
distribution of the sample mean will be
approximately normal, even if the distribution of
the population mean is not normal. This means
that we can use the normal distribution to
calculate a confidence interval for the sample
mean. (Refer the previous example)
Thanks!
Do you have any questions?