ESD EXAM

ESD.
86
Models, Data, Inference for Socio-Technical Systems
MIT, Spring 2007
Exam #2
Open Book, Open Notes, Individual Effort
1. (5 pts) A systems engineering textbook (Blanchard and Fabrycky) says “When the
system operating time is the Mean Time Between Failures (MTBF), the reliability is
37%.” Show that this is true for systems with constant failure rates.
2. (10 pts) The graph below is from Weibull's 1951 paper. Estimate the mean, median,
and the mode of the fatigue life. You do not need to be very precise in your estimates,
but you should explain clearly your approach and why it's valid. Discuss the
relationships among these the mean, median, and mode and what the relationships among
them imply about the distribution of the data.
Copyright © 1951 by ASME. Used with permission.

3. Given that variables z1, z2, and z3 are standardized, normally distributed, and
independent, how are the following random variables distributed? (name the distribution
and give the required parameters or else write the equation out in full):
a) (5 pts) z 2 + z 3
b) (5 pts) z2 + z3
2 2
z1
c) (5 pts) 0.1702 5.0223 2.3869
( z2 + z3 ) k
2 2
-0.6328 6.2121 3.4204
0.9821 4.662 4.0464
1.3432 4.5406 3.6269
-0.4473 3.8452 4.8308
2.1877 5.3442 3.257
1.9791 3.6864 4.2428
0.9132 5.658 1.174
1.4637 9.5435 2.8336
1.1773 4.3546 2.4748
4. (10 pts) Given the following commands: 0.4533 6.1233 -0.2493
1.6657 6.9879 3.1136
mu=[1 5 3]; -0.0869 2.2616 1.5618
sigma=[1 0 0; 0 2 0; 0 0 2]; 2.6525 3.3038 5.7646
0.7161 5.558 1.8655
n=15;
X = lhsnorm(mu,sigma,n)
These data are generated. The documentation says "X = lhsnorm(mu,sigma,n)

generates a Latin Hypercube sample X of size n from the multivariate normal distribution
with mean vector mu and covariance matrix sigma. What operations would you perform
to check that the sample actually is what the documentation claims it to be? (You don't
need to actually process the data, it's just provided in case you want to illustrate
concretely what you'd do).
5. (5 pts) You perform a linear regression of 100 data points (n=100). There are two
independent variables x1 and x2. The regression R2 is 0.72. Both β1 and β2 pass a t test
for significance. You decide to add the interaction x1x2 to the model. Select all the things
that cannot happen:
a) R2 decreases
c) The adjusted R2 decreases
d) All three coefficients β1, β2, and β12 fail the t test for significance
6. (10 pts) Provide a persuasive and rigorous defense of your answer to question #5
above. The answer need not be primarily implemented with symbolic mathematics, but it
could be.
7. The computer output below represents the results from a one way (single factor)
ANOVA (α=0.05). Each row represents a different treatment condition.
a) (3 pts) What does the “F crit” value represent?
b) (3 pts) What does the “F” value represent?
c) (3 pts) What are the most critical assumptions of this analysis?
d) (3 pts) Do you think any of the assumptions were violated?
e) (3 pts) What conclusion do you draw about the effect of the treatment? Briefly justify
your position.
Treatment Replicates
1 19 19 20 14 15 14
2 22 23 26 11 12 12
3 19 59 11 38 24 45
4 28 30 34 19 20 22
Anova: Single Factor
SUMMARY
Groups Count Sum Average Variance
1 6 101 16.83333 7.766667
2 6 106 17.66667 45.06667
3 6 196 32.66667 321.0667
4 6 153 25.5 36.7
ANOVA
Source of
Variation SS df MS F P-value F crit
Between
Groups 996.3333 3 332.1111 3.235374 0.043995 3.098393
Within Groups 2053 20 102.65
Total 3049.333 23
8. For the experimental plan and associated data in the Table below.
a) (5 pts) Sketch a factor effect plot.
b) (5 pts) Does this data strongly suggest that there are interactions among the factors?
Explain briefly.
Control Factors
Expt. A B C D response
No.
1 1 1 1 1 2.2
2 1 2 2 2 1.9
3 1 3 3 3 2.1
4 2 1 2 2 4
5 2 2 3 1 4.3
6 2 3 1 2 3.8
7 3 1 3 3 5.9
8 3 2 1 3 6.4
9 3 3 2 1 5.9
9. You run a 27 full factorial design. The response is the velocity of a paper helicopter
when it hits the ground when dropped from 9 feet. The standard deviation of the
experimental error is estimated to be 0.3 seconds since it's known the greatest
contribution to error is the response time in hitting the stopwatch.
a) (5 points) What is the standard deviation of a main effect estimate?
b) (5 points) Suggest three ways to reduce the uncertainty in the main effect estimate.
10. (10 points) You run a 27 full factorial design. You have the same situation as
problem #9 above, except now it's also known that one of the helicopters was not set to
the correct factor levels. Instead, it was set to the levels for some other helicopter in the
matrix experiment. Can you still characterize the uncertainty in your main effect
estimate?
If "yes", briefly describe a procedure would you suggest to estimate the uncertainty.
If "no", explain why not and describe what you'll do next.

ESD EXAM

Uploaded by

Copyright:

Available Formats

ESD EXAM

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ESD EXAM

Uploaded by

Copyright:

Available Formats

ESD.

Copyright © 1951 by ASME. Used with permission.

These data are generated. The documentation says "X = lhsnorm(mu,sigma,n)

a) (3 pts) What does the “F crit” value represent?

b) (3 pts) What does the “F” value represent?

c) (3 pts) What are the most critical assumptions of this analysis?

d) (3 pts) Do you think any of the assumptions were violated?

Anova: Single Factor

a) (5 pts) Sketch a factor effect plot.

a) (5 points) What is the standard deviation of a main effect estimate?

You might also like