Name:
QMBU 301 – Fall 2022
QUIZ 3B - Solution
Total time: 20 minutes
Altogether the quiz has a maximum of 35 points.
There are 4 pages including this page. The last two pages contain formulas and z-
table, respectively.
This is a closed-book, closed-notes exam.
You may use a simple calculator.
No laptops. No cell phones.
There is a partial credit for each question unless otherwise noted. Show your
work completely. If you don’t show your work, there will be no partial credits.
Do not be tempted to seek “inspiration” from a neighbor’s exam. Poor scores can
be overcome, but cheating will be dealt with harshly!
Please sign here to indicate that you have adhered to university policies regarding
ethical behavior in preparing for and completing this exam.
“I hereby certify that I have adhered to the university policies regarding ethical
behavior in preparing for and completing this exam.”
Signature:
1
Demand of a specific product in the first four weeks is given in the following table:
Week Demand MA(3) Absolute Error ES(0.4) Absolute Error
1 40 --------- --------- --------- ---------
2 37 --------- --------- 40 3
3 43 --------- --------- 38.8 4.2
4 36 40 4 40.48 4.48
5 --------- 38.67 --------- 38.69 ---------
MAE = 4 MAE = 3.89
1. (10 pts) Forecast the demand of the 5th week by MA(3) and ES(0.4).
(5pts) Under MA(3), 𝐹! = 38.67
(5pts) Under ES(0.4), 𝐹! = 38.69
2. (15 pts) Compute the mean absolute error (MAE) of MA(3) and ES(0.4). Among these two
forecasting methods, which one performs the best with respect to MAE in this example?
(5pts) Under MA(3), 𝑀𝐴𝐸 = 4
(5pts) Under ES(0.4), 𝑀𝐴𝐸 = (3 + 4.2 + 4.48)⁄3 = 3.89
(5pts) Because MAE under ES(0.4) is the smallest, ES(0.4) performs the best in this
example.
2
3. (10 pts) Construct an approximate 96% prediction interval (PI) for the demand in the 5th
week with the forecasting method performing the best with respect to MAE among MA(3)
and ES(0.4).
We will use ES(0.4) to construct a 96% PI for 𝑌! , which is 𝐹! ± 𝑧"."$ × 1.25 × 𝑀𝐴𝐸 where
𝑧"."$ solves the equality 𝑃(𝑍 ≤ 𝑧"."$ ) = 0.98. By z-table, 𝑧"."$ = 2.05 or 2.06. Then 96%
PI for 𝑌! is
• 38.69 ± 𝟐. 𝟎𝟓 × 1.25 × 3.89 ⟹ 38.69 ± 9.9767 ⟹ (28.7113, 48.6647)
• 38.69 ± 𝟐. 𝟎𝟔 × 1.25 × 3.89 ⟹ 38.69 ± 10.0253 ⟹ (28.6627, 48.7133)
Formulas and z-table
'
Under MA(𝑘), 𝐹%&' = ( (𝑌% + 𝑌%)' + ⋯ + 𝑌%)(&' ) for 𝑡 ≥ 𝑘
Under ES(𝛼),
• 𝐹$ = 𝑌'
• 𝐹%&' = 𝛼𝑌% + (1 − 𝛼)𝐹% for 𝑡 ≥ 2
'
𝑀𝐴𝐸 = * ∑*%+'|𝑌% − 𝐹% | where 𝑇 is the number of times the absolute error |𝑌% − 𝐹% | is
computed.
An approximate 100(1 − 𝛼)% PI for 𝑌%&' is 𝐹%&' ± 𝑧, ⁄$ × 1.25 × 𝑀𝐴𝐸 where 𝑧,⁄$ solves the
equality 𝑃L𝑍 ≤ 𝑧,⁄$ M = 1 − 𝛼 ⁄2
z-table is in the next page à
3
4
