ISEN 625 - Exam 2

How to deal with incomplete/nonexistent data regarding: -arrival times -service times: -mean available -range information available What do you need to be careful of?

-rate tables -non stationary arrivals/schedules -introduce initial variability (mean -/+ x% of mean) -uniform(min,max) triangular(min, midpoint, max) normal(midpoint, .25*range) -having potential negative values for service/arrival times

The Chi-Square test can be applied on a large set (> 50) of pseudo random numbers to test for A runs B uniformity C auto correlationship D independence E gaps

B

The consistency test for checking reasonableness of a model checks whether A small changes in input parameters leads to proportional changes in generated outputs. B similar runs yield similar results. C absurd conditions are occurring during simulation. D absurd inputs are producing absurd outputs. E structure of model is contradicting reality.

B

Which one of these is NOT used for testing of independence of pseudo random numbers? A Autocorrelation test B Frequency test C Runs above and below mean test D Gap test E Runs test

B

The number of degrees of freedom in a chi-square test depends on A Variance of data set B Number of classes formed C number of estimated parameters D sample size E Mean value of the data set

B & C

A series of pseudo random numbers must exhibit the properties of A positive correlationship B independence C negative correlationship D normality E uniformity

B & E

Which of the following concepts is NOT part of the reasonableness tests during validation? A Degeneracy B Continuity C Face Validation D Consistency E Absurd Conditions

C

4 steps of input analysis

1. Collect data from real system of interest 2. Identify probability distribution to represent input process 3. Choose the parameter(s) that determines the specific distribution 4. Evaluate the chosen distribution and estimated parameters

mersenne twister method has been shown to have a periodicity in the order of

10^6001

Conceptual validation attempts to answer the question: A Does the model adequately represent the real world system? B Have I built the model right? C Is there a deadlock in the model? D Does the ultimate user have confidence in the model's results? E Are the model generated behavior data characteristic of the real world system's behavioral data?

A

During validation, the process where every assumption and outcome is empirically tests and validated is known as A empiricism B rationalism C sensitivity analysis D face validation E degeneracy

A

If there is sufficient amount of available data, then the first task that you would perform as part of the input analysis process would be A to create a histogram. B to look for outliers. C to collect more data. D the goodness of fit tests. E normalize the data.

A

The Inverse Transform Technique relies on the use of A cumulative distribution function (CDF) B probability mass function (PMF) C probability distribution function (PDF) D empirical distributions E correlated random numbers

A

Which of the following is NOT part of the Verification process? A Establish a doubting frame of mind B Incorporate outside doubters C Model and Experment Walkthrough D Test Model Behavior E Test Runs

D

Which of the following is NOT classified as a common error that is likely to be encountered during the verification process? A Flow control B Initialization error C Overwriting variables D Blockage/deadlock E Behavior anomaly

E

While using a Linear Congruential Generator [ Xi+1 = (aXi + c) mod m ], the maximum period that you are ever going to achieve is 2^m m^2 m+1 m-1 m

E

Empirical distribution

no obvious fit with any specific distribution

Two assumptions for input analysis

data is available data is reliable and accurate

T/F Failing to reject an invalid model is a Type II error (or model user's risk).

true

T/F It is preferred to have intervals (or classes) with equal probabilities than equal width while applying the Chi-Square test on continuous data.

true

T/F The CDF of a continuous distribution can be obtained by integrating the probability distribution function (PDF) over the range of the random variable.

true

T/F The Chi-Square Test formalizes the concept of histogram-based approach of data analysis.

true

T/F The process of determining that the model operates as intended is called verification.

true

T/F The stress test is used during the verification process to create test cases that explore boundary conditions.

true

T/F While using a multiplicative congruential generator (c = 0) and m = 2b, you can obtain maximum periodicity P = m/4 if a = 3 + 8k or a = 5 + 8k (where k is an integer) AND X0 is odd.

true

T/F With a very large sample of data, it is possible that all candidate distributions are rejected while performing a goodness of fit test.

true

T/F the chi square test cannot be applied to a data set where parameters have not been estimated.

true

T/F Model debugging is an essential component of the validation process.

false

T/F Numbers generated from uniform distribution with a range of 0 - 10 are Random Numbers

false

T/F Pseudo random numbers generated by a method such as a Linear Congruential Generator will be continuous.

false

T/F The K-S Test cannot be applied to a data set where parameters have not been estimated.

false

T/F The K-S test determines the minimum deviation of the data set from the theoretical CDF of the underlying distribution.

false

T/F The Mersenne Twoster method for generatiing pseudo random numbers in Simio has approximately the same periodicity as the Linear Congruential Method with m = (2^31) - 1.

false

T/F The introduction of randomness or variability in the arrival or service time in a completely deterministic queueing system does not lead to system performance degradation (e.g. higher waiting times, larger time spent in system).

false

T/F When we divide the collected data into intervals of equal width, the choice of interval width has no effect on the shape of the histogram.

false

T/F While modeling service times you should use the random variates as they are generated, even if the random variate is equal to 0 or less than 0.

false

T/F While using an open tail distribution (e.g. Normal Distribution), it is impossible to generate a random variate that is less than zero.

false

A quantile-quantile (Q-Q) plot is used A to perform output analysis. B for visualizing the PDF of a distribution. C to generate random variates. D to evaluate the fit of a distribution. E as an alternative to the Chi-Square test.

D