Scribd
Upload
41 views3 pages

Management Science HW 6

This document outlines a group homework assignment in management science. It consists of 5 questions worth a total of 120 points. The questions involve time series forecasting using exponential smoothing, linear regression, and Holt-Winters' method on various datasets relating to sales, shipments, and power demand. Students are asked to write equations, make predictions, compare models, and interpret statistics to demonstrate their understanding of the relevant forecasting techniques.

Uploaded by

Danny Wong
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
41 views3 pages

Management Science HW 6

This document outlines a group homework assignment in management science. It consists of 5 questions worth a total of 120 points. The questions involve time series forecasting using exponential smoothing, linear regression, and Holt-Winters' method on various datasets relating to sales, shipments, and power demand. Students are asked to write equations, make predictions, compare models, and interpret statistics to demonstrate their understanding of the relevant forecasting techniques.

Uploaded by

Danny Wong
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Management Science HW 6

Group HW
Total scores of the assignment are 120pts. Thus, it is an opportunity for your group to
get more than 100 pts and make up for your prior hw grades. Good luck!

SmallBusiness.csv is the data of annual sales (in thousands) for a small business.
Q1. Create an exponential smoothing model.-21pts
(a) As it is a recursive function, write down the equations to show how F4 (forecast
of year 4) is derived from F1 using alpha = 0.9 and 0.01. -8pts
(b) prepare a line graph comparing the exponential smoothing predictions using
alpha = 0.9 and 0.01 against the original data. -8pts
(c) Is it correct that forecasts for the next two years are the same? Why or Why
not? -5pts
Q2. Use regression analysis to fit a linear trend model (time as the independent
variable).-21pts
(a) What is the estimated regression function? -5pts
(b) What is the value of R2? What is the meaning? -6pts
(c) prepare a line graph comparing the linear trend predictions against the original
data. -4pts
(d) If you would like to compare the model with the exponential smoothing
model, which model is better based on average of forecast error? Any
concerns? -6pts

Q3. Refer to the multiplicative seasonal effect in the lecture note. Please implement
the function and provide forecasts of 2003 using the TrendRegression.csv. -10pts

Q4. The time plot in Figure below shows the series of quarterly shipments (in million
dollars) of US household appliances between 1985-1989 (ApplicanceShipments.csv)-
31pts
(a) Which of the following methods would be suitable for forecasting this series if
applied to the raw data? (1) moving average (2) simple exponential smoothing (3)
Holt-Winter’s model-4pts
(b) Apply a moving average with window span w=4 to th data. Use all but the last
year as the training set. Create a time plot for the moving average series. Is the
forecast for Q1-1990 most likely to underestimate, overestimate or accuratley
estimate the acutal sales on Q1-1990? Explain.-8pts
(c) We now focus on forecasting beyond 1989. Use all but the last year as the training
set and the last four quarters as the validation set. First, fit a regression model to
sales with a linear trend and quarterly seasonality to the training data. Next, apply
Holt-Winter’s model with multiplicative seasonal effect to the training data. -12pts
(d) Following the above question (d), compute MAPE for the validation data for both
models. Which model would you prefer for forecasting Q1-1990?-7pts

Q5 Duque Power Company wants to develop a regression model to help predict its
daily peak power demand. This prediction is useful in determining how much
generating capacity needs to be available (or purchased from competitors) on a daily
basis. The daily peak power demand is influenced primarily by the weather and the
day of the week. The file Duque.csv contains data summarizing Duque’s daily peak
demand and maximum daily temperature during the month of July last year.-37pts
(a) Build a simple linear regression to predict peak power demand using maximum
daily temperature. What is the estimated regression equation? -8pts
(b) Build a multiple linear regression model to predict peak power demand using
maximum daily temperature and the day of the week as the independent
variables. (Note: You need to use dummy variables for the day of the week.)
What is the estimated regression model? What is the R^2 statistic of the model?
Interpret the meaning of the statistics. -12pts
(c) Prepare a line chart plotting the actual peak demand against the values predicted
by the two regression equations above. Which model is better based on the
diagram? Other than eyes scanning, what is the model statistic you would report
for the model comparison? -8pts
(d) Using the multiple regression model you developed, what is the predicted peak
power demand Duque should expect on a Wednesday in July when the daily high
temperature is forecast to be 94? Compute a 95% prediction interval for the
estimate, and explain the managerial implication of the interval to the Duque. -
9pts
Please turn in a pdf containing your group members, the code with clear comments,
concise answer and/or graphics. Use your Group_HW6 as the file name and upload
the file onto WM5. Note that TA have rights to deduct points if your documents/codes
are hard to follow

You might also like