Simulation set-up

21 important questions on Simulation set-up

What are, according to the slides, the four important steps of a simulation study?

1. Problem definition
2. Model construction
3. Experimental design and result analysis

What actions are undertaken in the problem definition phase?

Among others: Problem identification, goal setting, experimental factors, scope, level of detail, data collection and study planning.

Based on what factors is the scope/level of detail of the study determined?

1. The goal of the simulation study
2. The available time/budget
  • Higher grades + faster learning
  • Never study anything twice
  • 100% sure, 100% understanding
Discover Study Smart

Which three categories of data are explained in the lecture slides?

1. Available
2. Not available, but collectable
3. Not available and not collectable

How to overcome non-available and non-collectable data?

1. Experts opinion
2. Adjust project goals

What three sub-steps are there to model construction?

1. Determine model-structure
2. Implementation
3. Model validation

Why should you document your model construction?

1. For re-use of the model
2. In case multiple developers work on the model

What is the difference between validation, verification and credibility?

1. Validation ensures that the model is a good enough representation of the reality.
2. Verification ensures that the model represents the papermodel. (Do you model what you want to model?)
3. Credibility ensures that the decision maker accepts the models and results as being correct.

How can you verify the model?


-Use modules and test modules separately
-Let more than one person review the program (structured walk-through)
-Carry out preliminary runs and check whether the output and the effects of modifying parameters is plausible
-Tracing: list the events and system state variables option: stop program, modify a certain parameter to force a certain event to happen (e.g., arrival of a certain patient) and watch whether the intended effect occurs

-Run simplified model for which analytical results are available and compare output
-Observe animation
-Check input random generators
-Use built-in debugger to find and solve errors! (DEMO)

How can the validity and credibility of a model be increased?

1. Black box validation
2. Compare model with reality
3. Use animation
4. Ture test: Let experts guess which data is from the model/reality.

How can a model be compared with reality in a statistical way?

1. Inspection approach. Either by using the real-world input as input for the system and analyzing the outcomes, or compare summary statistics
2. Confidence interval approach

Why is a confidence interval preferred over a statistical test?

A statistical test will in most cases reject h0 and a confidence interval provides more information than the statistical test.

What is the meaning of a factor, in experimental design?

A factor is an input parameter or structural assumption.

What is the meaning of level, in experimental design?

The value of a certain factor.(e.g. #of machines)

What is the most important idea behind experimental design?

To decide in an efficient manner which configurations to simulate, before runs are made, in order to obtain desired output.

What is a drawback of the OFAT-method?

OFAT does not take interaction effects into account. The impact of a factor, may depend on the levels of the other factors.

How can one determine which factors have the greatest effect on a response?

1. Full case factorial design. This is only possible when a limited range of factor values are studied. Here the effects of all interactions between factors are studied.
2. 2k factorial design. Here one only considers two levels of each factor and their interactions.

What possible outcomes for main effects are there for two-way interaction effects in 2-factor design(22).

-No interaction
-Positive interaction
-Negative interaction

What are drawbacks of the 2k-factorial design?

1. Main effects are based on linear relations. This is obviously not always the case
2. Effects depend on the choice for the low and high-factor level. 
3. Number of experiments explodes with number of factors

What is 2k-p factorial design? What are its 'issues'?

It reduces the number of interaction effects that are studied by selecting a sub-set.
Challenges:
-It is hard to determine what subset of experiments to choose
- Two effects are confounded in a fractional factorial design if the formulas to calculate both effects are exactly the same. As a result we can only estimate the joint effect of two effects.

What is the issue with CRN that are used to decrease the width of the Confidence Interval for Interaction Effects?

Due to correlation of the outcomes, the covariance plays a role. This may increase the variance, and therefore the width of the confidence interval.

The question on the page originate from the summary of the following study material:

  • A unique study and practice tool
  • Never study anything twice again
  • Get the grades you hope for
  • 100% sure, 100% understanding
Remember faster, study better. Scientifically proven.
Trustpilot Logo