ch17 Solutions

Chapter 17: Forecasting
Textbook Exercises:
1. Consider the following time series data.
Week 1 2 3 4 5 6
Value 18 13 16 11 17 14
Using the naive method (most recent value) as the forecast for the next week, compute the following
measures of forecast accuracy.
a. Mean absolute error.
b. Mean squared error.
c. Mean absolute percentage error.
d. What is the forecast for week 7?
2. Refer to the time series data in exercise 1. Using the average of all the historical data as a forecast for
the next period, compute the following measures of forecast accuracy.
a. Mean absolute error.
b. Mean squared error.
c. Mean absolute percentage error.
d. What is the forecast for week 7?
3. Exercises 1 and 2 used different forecasting methods. Which method appears to provide the more
accurate forecasts for the historical data? Explain.

Month 1 2 3 4 5 6 7
Value 24 13 20 12 19 23 15
a. Compute MSE using the most recent value as the forecast for the next period. What is the
forecast for month 8?
b. Compute MSE using the average of all the data available as the forecast for the next period. What
is the forecast for month 8?
c. Which method appears to provide the better forecast?
Week 1 2 3 4 5 6
Value 18 13 16 11 17 14
a. Construct a time series plot. What type of pattern exists in the data?
b. Develop the three-week moving average forecasts for this time series. Compute MSE and a
forecast for week 7.
c. Use α = 0.2 to compute the exponential smoothing forecasts for the time series. Compute MSE
and a forecast for week 7.
d. Compare the three-week moving average approach with the exponential smoothing approach
using α = 0.2. Which appears to provide more accurate forecasts based on MSE? Explain.
e. Use a smoothing constant of α = 0.4 to compute the exponential smoothing forecasts. Does a
smoothing constant of 0.2 or 0.4 appear to provide more accurate forecasts based on MSE?
Explain.
Month 1 2 3 4 5 6 7
Value 24 13 20 12 19 23 15
Construct a time series plot. What type of pattern exists in the data?
a. Develop the three-week moving average forecasts for this time series. Compute MSE and a
forecast for week 8.
b. Use α = 0.2 to compute the exponential smoothing forecasts for the time series. Compute MSE
and a forecast for week 8.
c. Compare the three-week moving average approach with the exponential smoothing approach
using α = 0.2. Which appears to provide more accurate forecasts based on MSE?
d. Use a smoothing constant of α = 0.4 to compute the exponential smoothing forecasts. Does a
smoothing constant of 0.2 or 0.4 appear to provide more accurate forecasts based on MSE?
Explain.
7. Refer to the petrol sales time series data in Table 17.1.
a. Compute four-week and five-week moving averages for the time series.
b. Compute the MSE for the four-week and five-week moving average forecasts.
c. What appears to be the best number of weeks of past data (three, four, or five) to use in the
moving average computation? Recall that MSE for the three-week moving average is 10.22.
8. Refer again to the petrol sales time series data in Table 17.1.
a. Using a weight of 1/2 for the most recent observation, 1/3 for the second most recent observation,
and 1/6 for third most recent observation, compute a three-week weighted moving average for the
time series.
b. Compute the MSE for the weighted moving average in part (a). Do you prefer this weighted
moving average to the unweighted moving average? Remember that the MSE for the unweighted
moving average is 10.22.
c. Suppose you are allowed to choose any weights as long as they sum to 1. Could you always find
a set of weights that would make the MSE at least as small for a weighted moving average than
for an unweighted moving average? Why or why not?
9. With the petrol time series data from Table 17.1, show the exponential smoothing forecasts using α =
0.1.
a. Applying the MSE measure of forecast accuracy, would you prefer a smoothing constant of α =
0.1 or α = 0.2 for the petrol sales time series?
b. Are the results the same if you apply MAE as the measure of accuracy?
c. What are the results if MAPE is used?
10. With a smoothing constant of α = 0.2, equation (17.2) shows that the forecast for week 13 of the
petrol sales data from Table 17.1 is given by F13 = 0.2Y12 + 0.8F12. However, the forecast for week 12
is given by F12 = 0.2Y11 + 0.8F11. Thus, we could combine these two results to show that the forecast
for week 13 can be written
F13 = 0.2Y12 + 0.8(0.2Y11 + 0.8F11) = 0.2Y12 + 0.16Y11 + 0.64F11
a. Making use of the fact that F11 = 0.2Y10 + 0.8F10 (and similarly for F10 and F9), continue to expand
the expression for F13 until it is written in terms of the past data values Y12, Y11, Y10, Y9, Y8, and the
forecast for period 8.
b. Refer to the coefficients or weights for the past values Y12, Y11, Y10, Y9, Y8. What observation can
you make about how exponential smoothing weights past data values in arriving at new
forecasts? Compare this weighting pattern with the weighting pattern of the moving averages
method.
11. For SIS Cargo Services in Dubai, the monthly percentages of all shipments received on time over the
past 12 months are 80, 82, 84, 83, 83, 84, 85, 84, 82, 83, 84, and 83.
b. Compare the three-month moving average approach with the exponential smoothing approach for
α = 0.2. Which provides more accurate forecasts using MSE as the measure of forecast accuracy?
c. What is the forecast for next month?
12. The values of Austrian building contracts (in millions of euros) for a 12-month period follow.
240 350 230 260 280 320 220 310 240 310 240 230
b. Compare the three-month moving average approach with the exponential smoothing forecast
using α = 0.2. Which provides more accurate forecasts based on MSE?
c. What is the forecast for the next month?
13. The following data represent indices for the merchandise imports by broad economic category
for New Zealand from 2013 to 2017:
2013 Jun 1,177 2016 Mar 945

Sep 1,202 Jun 972
Dec 1,173 Sep 969
2014 Mar 1,162 Dec 971
Jun 1,133 2017 Mar 1,013
Sep 1,149
Dec 1,136
2015 Mar 1,022
Jun 1,041
Sep 1,118
Dec 1,035
a. Compute three- and four-quarter moving averages for this time series. Which moving average
provides the better forecast for the second quarter of 2017?
b. Plot the data. Do you think the exponential smoothing model would be appropriate for forecasting
in this case?
t 1 2 3 4 5
Yt 6 11 9 14 15
b. Develop the linear trend equation for this time series.
c. What is the forecast for t = 6?
15 Refer to the time series in exercise 14. Use Holt’s linear exponential smoothing method with α =
0.3 and β = 0.5 to develop a forecast for t = 6.
16. Consider the following time series.
t 1 2 3 4 5 6 7
Yt 120 110 100 96 94 92 88
a. Construct a time series plot. What type of pattern

exists in the data?
b. Develop the linear trend equation for this time series.
c. What is the forecast for t =8?
17. Consider the following time series.
t 1 2 3 4 5 6 7
Yt 82 60 44 35 30 29 35
b. Using Minitab or Excel, develop the quadratic trend equation for the time series.
c. What is the forecast for t = 8?
18. The number of social network users worldwide (www.statista.com) from 2010 to 2018 was as
follows:
Year Period Users (billions)

2010 1 0.97
2011 2 1.22
2012 3 1.40
2013 4 1.59
2014 5 1.91
2015 6 2.14
2016 7 2.28
2017 8 2.46
2018 9 2.62
a. Construct a time series plot. What type of pattern exists?

b. Using Minitab or Excel, develop a quadratic trend.
19. Numbers of overseas visitors to Ireland (’000s) estimated by Failte Ireland for the years 2009–17
are as follows:
2009 2010 2011 2012 2013 2014 2015 2016 2017

6,578 5,945 6,240 6,286 6,686 7,105 8,036 8,742 9,023
a. Graph the data and assess its suitability for linear trend projection.
b. Use a linear trend projection to forecast this time series for 2018–20.
20.
GDP (USD billion) for 2008–17 are tabulated below (tradingeconomics.com/singapore/gdp):
Year USD (billions)
2008 192.23
2009 192.41
2010 236.42
2011 275.97
2012 290.67
2013 304.45
2014 311.54
2015 304.10
2016 309.76
2017 323.91
a. Graph this time series. Does a linear trend appear to be present?

b. Develop a linear trend equation for this time series.
c. Use the trend equation to estimate the GDP for the years 2018–20.
21. Monthly CD sales (in units) for a new string quartet over an 18 month period are as follows:
Month Sales
1 293
2 283
3 322
4 355
5 346
6 379
7 381
8 431
9 424
10 433
11 470
12 481
13 549
14 544
15 601
16 587
17 644
18 660
a. Using Holt’s model, obtain forecasts for sales for each of the next three months. (Take for
weighting parameters α = 0.3, β = 0.4.)
b. Justify the starting values for S and b that you use in your modelling.
c. By plotting fitted against actual values, comment on the likely quality of your forecasts.
22 Consider the following time series.
Quarter Year 1 Year 2 Year 3

1 71 68 62
2 49 41 51
3 58 60 53
4 78 81 72
b. Use the following dummy variables to develop an estimated regression equation to account for
seasonal effects in the data: Qtr1 = 1 if Quarter 1, 0 otherwise; Qtr2 = 1 if Quarter 2, 0 otherwise;
Qtr3 = 1 if Quarter 3, 0 otherwise.
c. Compute the quarterly forecasts for next year.
23 Consider the following time series data.

1 4 6 7
2 2 3 6
3 3 5 6
4 5 7 8
any seasonal and linear trend effects in the data: Qtr1 = 1 if Quarter 1, 0 otherwise; Qtr2 = 1 if
Quarter 2, 0 otherwise; Qtr3 = 1 if Quarter 3, 0 otherwise.
24 The quarterly sales data (number of copies sold) for a college textbook over the past three years
follow.

1 1690 1800 1850
2 940 900 1100
3 2625 2900 2930
4 2500 2360 2615
any seasonal effects in the data: Qtr1 = 1 if Quarter 1, 0 otherwise; Qtr2 = 1 if Quarter 2, 0
otherwise; Qtr3 = 1 if Quarter 3, 0 otherwise.
d. Let t = 1 to refer to the observation in quarter 1 of year 1; t = 2 to refer to the observation in
quarter 2 of year 1; ... and t = 12 to refer to the observation in quarter 4 of year 3. Using the
dummy variables defined in part (b) and t, develop an estimated regression equation to account
for seasonal effects and any linear trend in the time series. Based upon the seasonal effects in the
data and linear trend, compute the quarterly forecasts for next year.
25 Air pollution control specialists in northern Poland monitor the amount of ozone, carbon dioxide,
and nitrogen dioxide in the air on an hourly basis. The hourly time series data exhibit seasonality,
with the levels of pollutants showing patterns that vary over the hours in the day. On July 15, 16,
and 17, the following levels of nitrogen dioxide were observed for the 12 hours from 6:00 A. M. to
6:00 P. M.
July 15: 25 28 35 50 60 60 40 35 30 25 25 20
July 16: 28 30 35 48 60 65 50 40 35 25 20 20
July 17: 35 42 45 70 72 75 60 45 40 25 25 25
the seasonal effects in the data.
Hour1 = 1 if the reading was made between 6:00 A.M. and 7:00 A.M.; 0 otherwise
Hour2 = 1 if if the reading was made between 7:00 A.M. and 8:00 A.M.; 0 otherwise
.
.
.
Hour11 = 1 if the reading was made between 4:00 P.M. and 5:00 P.M., 0 otherwise.
Note that when the values of the 11 dummy variables are equal to 0, the observation corresponds
to the 5:00 P.M. to 6:00 P.M. hour.
c. Using the estimated regression equation developed in part (a), compute estimates of the levels of
nitrogen dioxide for July 18.
d. Let t = 1 to refer to the observation in hour 1 on July 15; t = 2 to refer to the observation in hour 2
of July 15;0... and t = 36 to refer to the observation in hour 12 of July 17. Using the dummy
variables defined in part (b) and t, develop an estimated regression equation to account for
seasonal effects and any linear trend in the time series. Based upon the seasonal effects in the data
and linear trend, compute estimates of the levels of nitrogen dioxide for July 18.
26 Consider the following time series data.

1 4 6 7
2 2 3 6
3 3 5 6
4 5 7 8
b. Show the four-quarter and centred moving average values for this time series.
c. Compute seasonal indices and adjusted seasonal indices for the four quarters.
27 Refer to exercise 26.

a. Deseasonalize the time series using the adjusted seasonal indices computed in part (c) of exercise
35.
b. Using Minitab or Excel, compute the linear trend regression equation for the deseasonalized data.
c. Compute the deseasonalized quarterly trend forecast for Year 4.
d. Use the seasonal indices to adjust the deseasonalized trend forecasts computed in part (c).
28 The quarterly sales data (number of copies sold) for a college textbook over the past three years
follow.

1 1690 1800 1850
2 940 900 1100
3 2625 2900 2930
4 2500 2360 2615
a. Show the four-quarter and centred moving average values for this time series.
b. Compute the seasonal and adjusted seasonal indices for the four quarters.
c. When is the largest seasonal index? Does this result appear reasonable? Explain.
d. Deseasonalize the time series.
e. Compute the linear trend equation for the deseasonalized data and forecast sales using the linear
trend equation.
f. Adjust the linear trend forecasts using the adjusted seasonal indices computed in part (b).
29 Quarterly sales data for the number of houses sold over the past four years or so by a national
chain are as follows:
Year Ql Q2 Q3 Q4
1 200 212 229 207
2 195 204 216 202
3 201 209 221 205
4 208 217 231 213
5 218
a. Decompose the series into trend, seasonal and random components using a multiplicative model.
b. Hence derive forecasts of the number of houses that will be sold in the next four quarters.
c. Comment on the quality of your modelling results.
30 The following table shows the number of passengers per quarter (in thousands) who flew with
MBI Junior for the first quarter of this year and the three years preceding:
Year Q1 Q2 Q3 Q4
1 44 92 156 68
2 60 112 180 80
3 64 124 200 104
4 76
a. Decompose the series into trend, seasonal and random components using an additive
model.
b. Hence derive forecasts of the passenger numbers in the next four quarters.
c. Comment on the quality of your modelling.
31.
A public swimming pool records the following quarterly attendances (in thousands)
over a four year period. Using a (multiplicative) decomposition model, forecast
sales for the next four quarters in the series.
Year Quarter Attendances Year Quarter Attendances
1 1 5.0 3 1 6.3
2 3.8 2 4.8
3 3.2 3 4.6
4 4.6 4 5.9
2 1 5.4 4 1 7.4
2 4.7 2 5.6
3 4.7 3 4.8
4 6.1 4 6.8
Comment on the quality of your modelling.

Textbook Exercises Solutions:

1. The following table shows the calculations for parts a, b, and c.
Week Time-Series Forecast Forecast Error Absolute Value Squared Forecast Percentage Absolute Value of
Value of Forecast Error Error Percentage Error
Error
1 18
2 13 18 –5 5 25 –38.46 38.46
3 16 13 3 3 9 18.75 18.75
4 11 16 –5 5 25 –45.45 45.45
5 17 11 6 6 36 35.29 35.29
6 14 17 –3 3 9 –21.43 21.43
Totals 22 104 –51.30 159.38
a. MAE = 22/5 = 4.4
b. MSE = 104/5 = 20.8
c. MAPE = 159.38/5 = 31.88
d. Forecast for week 7 is 14
2. The following table shows the calculations for parts a, b, and c.
Week Time-Series Forecast Forecast Error Absolute Value Squared Forecast Percentage Absolute Value of
Value of Forecast Error Error Percentage Error
Error
1 18
2 13 18.00 –5.00 5.00 25.00 –38.46 38.46
3 16 15.50 0.50 0.50 0.25 3.13 3.13
4 11 15.67 –4.67 4.67 21.81 –42.45 42.45
5 17 14.50 2.50 2.50 6.25 14.71 14.71
6 14 15.00 –1.00 1.00 1.00 –7.14 7.14
Totals 13.67 54.31 –70.21 105.86
a. MAE = 13.67/5 = 2.73
b. MSE = 54.31/5 = 10.86
c. MAPE = 105.89/5 = 21.18
d. Forecast for week 7 is (18 + 13 + 16 + 11 + 17 + 14) / 6 = 14.83
3. The following table shows the measures of forecast error for both methods.
Exercise 1 Exercise 2
MAE 4.40 2.73
MSE 20.80 10.86
MAPE 31.88 21.18
For each measure of forecast accuracy, the average of all historical data provided more accurate
forecasts than simply using the most recent value.
4. a.
Month Time-Series Forecast Forecast Error Squared Forecast
Value Error
1 24
2 13 24 –11 121
3 20 13 7 49
4 12 20 –8 64
5 19 12 7 49
6 23 19 4 16
7 15 23 –8 64
Total 363
MSE = 363/6 = 60.5

Forecast for month 8 = 15.
b.
Week Time-Series Forecast Forecast Error Squared Forecast
Value Error
1 24
2 13 24.00 –11.00 121.00
3 20 18.50 1.50 2.25
4 12 19.00 –7.00 49.00
5 19 17.25 1.75 3.06
6 23 17.60 5.40 29.16
7 15 18.50 –3.50 12.25
Total 216.72
MSE = 216.72/6 = 36.12
Forecast for month 8 = (24 + 13 + 20 + 12 + 19 + 23 + 15) / 7 = 18.
c. The average of all the previous values is better because MSE is smaller.
5. a.
The data appear to follow a horizontal pattern.

b. Three-Week Moving Average
Week Value Forecast Forecast Error Squared Forecast
Error
1 18
2 13
3 16
4 11 15.67 –4.67 21.78
5 17 13.33 3.67 13.44
6 14 14.67 –0.67 0.44
Total 35.67
MSE = 35.67/3 = 11.89
The forecast for week 7 = (11 + 17 + 14) / 3 = 14.
c. Smoothing constant = 0.2
Value Error
1 18
2 13 18.00 –5.00 25.00
3 16 17.00 –1.00 1.00
4 11 16.80 –5.80 33.64
5 17 15.64 1.36 1.85
6 14 15.91 –1.91 3.66
Total 65.15
MSE = 65.15/5 = 13.03

The forecast for week 7 is 0.2(14) + (1 – 0.2)15.91 = 15.53.
d. The three-week moving average provides a better forecast because it has a smaller MSE.
e. Smoothing constant = 0.4
Value Error
1 18
2 13 18.00 –5.00 25.00
3 16 16.00 0.00 0.00
4 11 16.00 –5.00 25.00
5 17 14.00 3.00 9.00
6 14 15.20 –1.20 1.44
Total 60.44
MSE = 60.44/5 = 12.09
The exponential smoothing forecast using α = 0.4 provides a better forecast than the exponential
smoothing forecast using α = 0.2 because it has a smaller MSE.
6. a.
Three-Week Moving Average
Value Error
1 24
2 13
3 20
4 12 19.00 –7.00 49.00
5 19 15.00 4.00 16.00
6 23 17.00 6.00 36.00
7 15 18.00 –3.00 9.00
Total 110.00
MSE = 110/4 = 27.5.

The forecast for week 8 = (19 + 23 + 15) / 3 = 19.
b. Smoothing constant = 0.2

Value Error
1 24
2 13 24.00 –11.00 121.00
3 20 21.80 –1.80 3.24
4 12 21.44 –9.44 89.11
5 19 19.55 –0.55 0.30
6 23 19.44 3.56 12.66
7 15 20.15 –5.15 26.56
Total 252.87
MSE = 252.87/6 = 42.15
The forecast for week 8 is 0.2(15) + (1 – 0.2)20.15 = 19.12.
c. The three-week moving average provides a better forecast because it has a smaller MSE.
d. Smoothing constant = 0.4.
Week Time-Series Forecast Forecast Error Squared Value of
Value Forecast Error
1 24
2 13 24.00 –11.00 121.00
3 20 19.60 0.40 0.16
4 12 19.76 –7.76 60.22
5 19 16.66 2.34 5.49
6 23 17.59 5.41 29.23
7 15 19.76 –4.76 22.62
Total 238.72
MSE = 238.72/6 = 39.79
The exponential smoothing forecast using α = 0.4 provides a better forecast than the exponential
smoothing forecast using α = 0.2 because it has a smaller MSE.
7. a.
Week Time-Series Value Four-Week Moving (Error)2 Five-Week Moving (Error)2
Average Forecast Average Forecast
1 17
2 21
3 19
4 23
5 18 20.00 4.00
6 16 20.25 18.06 19.60 12.96
7 20 19.00 1.00 19.40 0.36
8 18 19.25 1.56 19.20 1.44
9 22 18.00 16.00 19.00 9.00
10 20 19.00 1.00 18.80 1.44
11 15 20.00 25.00 19.20 17.64
12 22 18.75 10.56 19.00 9.00
Totals 77.18 51.84
b. MSE(4-week) = 77.18 / 8 = 9.65
MSE(5-week) = 51.84 / 7 = 7.41
c. For the limited data provided, the five-week moving average provides the smallest MSE.
8. a.
Week Time-Series Weighted Moving Forecast Error (Error)2
Value Average Forecast
1 17
2 21
3 19
4 23 19.33 3.67 13.47
5 18 21.33 –3.33 11.09
6 16 19.83 –3.83 14.67
7 20 17.83 2.17 4.71
8 18 18.33 –0.33 0.11
9 22 18.33 3.67 13.47
10 20 20.33 –0.33 0.11
11 15 20.33 –5.33 28.41
12 22 17.83 4.17 17.39
Total 103.43
b. MSE = 103.43 / 9 = 11.49
Prefer the unweighted moving average here; it has a smaller MSE.
c. You could always find a weighted moving average at least as good as the unweighted one.
The unweighted moving average is a special case of the weighted ones where the weights are
equal.
9. The following tables show the calculations for α = 0.1.
Week Time-Series Forecast Forecast Error Absolute Value Squared Percentage Absolute Value of
Value of Forecast Forecast Error Error Percentage Error
Error
1 17
2 21 17.00 4.00 4.00 16.00 19.05 19.05
3 19 17.40 1.60 1.60 2.56 8.42 8.42
4 23 17.56 5.44 5.44 29.59 23.65 23.65
5 18 18.10 –0.10 0.10 0.01 –0.56 0.56
6 16 18.09 –2.09 2.09 4.37 –13.06 13.06
7 20 17.88 2.12 2.12 4.49 10.60 10.60
8 18 18.10 –0.10 0.10 0.01 –0.56 0.56
9 22 18.09 3.91 3.91 15.29 17.77 17.77
10 20 18.48 1.52 1.52 2.31 7.60 7.60
11 15 18.63 –3.63 3.63 13.18 –24.20 24.20
12 22 18.27 3.73 3.73 13.91 16.95 16.95
Totals 28.24 101.72 65.67 142.42
The following tables show the calculations for α = 0.2.

Week Time-Series Forecast Forecast Absolute Value of Squared Percentage Absolute Value of
Value Error Forecast Error Forecast Error Error Percentage Error
1 17
2 21 17.00 4.00 4.00 16.00 19.05 19.05
3 19 17.80 1.20 1.20 1.44 6.32 6.32
4 23 18.04 4.96 4.96 24.60 21.57 21.57
5 18 19.03 –1.03 1.03 1.06 –5.72 5.72
6 16 18.83 –2.83 2.83 8.01 –17.69 17.69
7 20 18.26 1.74 1.74 3.03 8.70 8.70
8 18 18.61 –0.61 0.61 0.37 –3.39 3.39
9 22 18.49 3.51 3.51 12.32 15.95 15.95
10 20 19.19 0.81 0.81 0.66 4.05 4.05
11 15 19.35 –4.35 4.35 18.92 –29.00 29.00
12 22 18.48 3.52 3.52 12.39 16.00 16.00
Totals 28.56 98.80 35.84 147.44
a. MSE for α = 0.1 = 101.72/11 = 9.25
MSE for α = 0.2 = 98.80/11 = 8.98
α = 0.2 provides more accurate forecasts based on MSE.
b. MAE for α = 0.1 = 28.24/11 = 2.57
MAE for α = 0.2 = 28.56/11 = 2.60
α = 0.1 provides more accurate forecasts based on MAE, but they are
extremely close.
c. MAPE for α = 0.1 = 142.42/11 = 12.95%
MAPE for α = 0.2 = 147.44/11 = 13.40%
α = 0.1 provides more accurate forecasts based on MAPE.
10. a.
F13 = 0.2Y12 + 0.16Y11 + 0.64(0.2Y10 + 0.8F10) = 0.2Y12 + 0.16Y11 + 0.128Y10 +
0.512F10
F13 = 0.2Y12 + 0.16Y11 + 0.128Y10 + 0.512(0.2Y9 + 0.8F9) = 0.2Y12 + 0.16Y11 +
0.128Y10 + 0.1024Y9 + 0.4096F9
F13 = 0.2Y12 + 0.16Y11 + 0.128Y10 + 0.1024Y9 + 0.4096(0.2Y8 + 0.8F8) = 0.2Y12
+ 0.16Y11 + 0.128Y10 + 0.1024Y9 + 0.08192Y8 + 0.32768F8
b. The more recent data receive the greater weight or importance in determining the
forecast. The moving averages method weights the last n data values equally in
determining the forecast.
11. a.
The first two time-series values may be an indication that the time series has shifted
to a new higher level as shown by the remaining 10 values. Overall, however, the
time-series plot exhibits a horizontal pattern.
b.
Moving Average Plot for C1

Variable
85
Actual
Fits
Moving Average
84
Length 3
Accuracy Measures
MAPE 0.97761
83
MAD 0.81481
MSD 1.23457
C1
82
81
80
1 2 3 4 5 6 7 8 9 10 11 12
Index
Smoothing Plot for C1

Single Exponential Method
85 Variable
Actual
Fits
84 Smoothing Constant
α 0.2
Accuracy Measures
83 MAPE 1.2562
MAD 1.04141
C1
MSD 1.72065
82
81
80
1 2 3 4 5 6 7 8 9 10 11 12
Index
From the plots above, it can be seen the MA model has an MSE (MSD)
of 1.23 compared 1.72 for the exponential smoothing model. Thus the MA
model would be preferred.
c. Forecasts
Period Forecast Lower Upper

13 83.3333 81.1556 85.5111
12. a.
b.
Month Time-Series Three-Month Moving (Error)2 α = 0.2 (Error)2
Value Average Forecast Forecast
1 240
2 350 240.00 12,100.00
3 230 262.00 1,024.00
4 260 273.33 177.69 255.60 19.36
5 280 280.00 0.00 256.48 553.19
6 320 256.67 4,010.69 261.18 3,459.79
7 220 286.67 4,444.89 272.95 2,803.70
8 310 273.33 1,344.69 262.36 2,269.57
9 240 283.33 1,877.49 271.89 1,016.97
10 310 256.67 2,844.09 265.51 1,979.36
11 240 286.67 2,178.09 274.41 1,184.05
12 230 263.33 1,110.89 267.53 1,408.50
Totals 17,988.52 27,818.49
MSE(3-month) = 17,988.52 / 9 = 1998.72
MSE(α = 0.2) = 27,818.49 / 11 = 2528.95
Based on the preceding MSE values, the three-month moving averages appear better.
However, exponential smoothing was penalized by including month 2, which was
difficult for any method to forecast. Using only the errors for months 4 to 12, the
MSE for exponential smoothing is:
MSE(α = 0.2) = 14,694.49 / 9 = 1632.72
Thus, exponential smoothing was better considering months 4 to 12.
c. Using exponential smoothing,
F13 = α Y12 + (1 – α)F12 = 0.20(230) + 0.80(267.53) = 260
13. a. MA3 results
Moving Average Plot for NZ merchandise imports

Variable
1200
Actual
Fits
Forecasts
1150 95.0% PI
NZ merchandise imports
Moving Average
1100 Length 3
Accuracy Measures
MAPE 2.66
1050 MAD 27.43
MSD 1496.59
1000
950
900
2 4 6 8 10 12 14 16
Index
MA4 results
Moving Average Plot for NZ merchandise imports
Variable
1200
Actual
Fits
Forecasts
1150
NZ merchandise imports 95.0% PI
Moving Average
1100 Length 4
Accuracy Measures
1050 MAPE 2.85
MAD 29.28
MSD 1558.61
1000
950
900
2 4 6 8 10 12 14 16
Index
Based on diagnostics (accuracy measures) provided, the 3-point moving average is

marginally better on all counts.
b.
1200
1150
1100
Index
1050
1000
950
0 5 10 15 20
Quarter
.
To be able to apply the simple exponential smoothing model we need to assume the data
are stationary. From the plot above, this does not appear to be the case.
14. a.
The time-series plot shows a linear trend.
b.
c.
15.
16. a.
The time-series plot shows a linear trend.
b.
c.
17. a.
The time-series plot exhibits a curvilinear trend.

b. The linear trend equation is =107.857 – 28.9881 t + 2.65476 r2.
c. =107.857 – 28.9881(8) +2.65476 = 45.86
18. a.
A very faintly non-linear pattern e.g. quadratic in evidence here.
b.
Trend Analysis Plot for Social media users

Quadratic Trend Model
Yt = 0.6962 + 0.2621×t - 0.00516×t^2
2.8 Variable
Actual
2.6
Fits
2.4 Accuracy Measures
Social media users
MAPE 1.74504
2.2
MAD 0.0284
2.0 MSD 0.00131
1.8
1.6
1.4
1.2
1.0
1 2 3 4 5 6 7 8 9
Index
Quadratic trend results above represent almost a perfect fit.

Minitab’s S-curve option has an MSD of 2513.39 so marginally fits the data
better but this quadratic approximation looks quite convincing
historically.
19.
a.
Index = 1 corresponds with 2001
2 …. 2002 etc
From the graph of the data below, linear trend projection does not look a particularly
strong option.
Trend Analysis Plot for Visitors to Ireland

Linear Trend Model
Yt = 5301 + 376.4×t
10000 Variable
Actual
Fits
Forecasts
9000
Accuracy Measures
Visitors to Ireland
MAPE 6
MAD 393
8000 MSD 210022
7000
6000
1 2 3 4 5 6 7 8 9 10 11 12
Index
b. Nevertheless, from EXCEL
Equation for linear trend: Tt = 5301 + 376.4xt
Hence forecasts are as follows:
T2018 = 9062 when xt = 10
T2019 = 9438 when xt = 11
T2020 = 9814 when xt = 12

20. a. From the graph below, a linear trend is not strongly evident.
Trend Analysis Plot for Singapore GDP

Linear Trend Model
Yt = 192.0 + 14.94×t
400 Variable
Actual
Fits
Forecasts
350
Accuracy Measures
Singapore GDP
MAPE 6.565
MAD 17.209
300 MSD 362.965
250
200
1 2 3 4 5 6 7 8 9 10 11 12 13
Index
b.
Nevertheless, from EXCEL
Equation for linear trend: Tt =192.0 + 14.94xt
c. Hence forecasts are as follows:
T2018 = 356 when xt = 11
T2019 = 371 when xt = 12
T2020 = 386 when xt = 13
21. a.
Holt's two-parameter exponential smoothing
alpha = 0.3
beta= 0.4
Month Yt Lt bt Ft
1 293 293.00 21.60
2 283 305.12 17.81 314.60
3 322 322.65 17.70 322.93
4 355 344.74 19.46 340.35
5 346 358.74 17.27 364.20
6 379 376.91 17.63 376.01
7 381 390.48 16.01 394.54
8 431 413.84 18.95 406.48
9 424 430.15 17.89 432.79
10 433 443.53 16.09 448.04
11 470 462.73 17.33 459.62
12 481 480.35 17.45 480.07
13 549 513.16 23.59 497.79
14 544 538.92 24.46 536.75
15 601 574.67 28.98 563.38
16 587 598.65 26.98 603.64
17 644 631.14 29.18 625.63
18 660 660.23 29.14 660.32
19 689.37
20 718.51
21 747.66
MSE = 438.53
b. L1 = 293 = Y1 and b1
c. The modelling appears quite effective and the forecasts therefore fairly convincing.
22 a. The time series plot shows a horizontal pattern, but there is a seasonal pattern in the
data; for instance, in each year the lowest value occurs in quarter 2 and the highest
value occurs in quarter 4
b. A portion of the Minitab regression output is shown;

The regression equation is
Value = 77.0 – 10.0 Qtr1 – 30.0
Qtr2 – 20.0 Qtr3
c. The quarterly forecasts for next year are as follows:
Quarter 1 forecast = 77.0 – 10.0(1) – 30.0(0) – 20.0(0) = 67
Quarter 3 forecast = 77.0 – 10.0(0) – 30.0(0) –20.0(1) = 57
23 a.
Four-Quarter Centred
Year Quarter Yt Moving Average Moving Average
1 1 4
2 2
3.50
3 3 3.750
4.00
4 5 4.125
4.25
2 1 6 4.500
4.75
2 3 5.000
5.25
3 5 5.375
5.50
4 7 5.875
6.25
3 1 7 6.375
6.50
2 6 6.625
6.75
3 6
4 8
b.
Centred Seasonal-Irregular
Year Quarter Yt Moving Average Component
1 1 4
2 2
3 3 3.750 0.8000
4 5 4.125 1.2121
2 1 6 4.500 1.3333
2 3 5.000 0.6000
3 5 5.375 0.9302
4 7 5.875 1.1915
3 1 7 6.375 1.0980
2 6 6.625 0.9057
3 6
4 8
c.
Seasonal-Irregular Adjusted Seasonal
Quarter Component Values Seasonal Index Index
1 1.3333,1.0980 1.2157 1.2050

2 0.60000,.9057 0.7529 0.7463
3 0.80000,.9032 0.8651 0.8675
4 1.2121,1.1915 1.2018 1.1912
4.0355
Note: Adjustment for seasonal index = 4.000 / 4.0355 = 0.9912
24. a.
There appears to be a seasonal pattern in the data and perhaps a moderate upward
linear trend.
b. The regression output follows.
Regression Equation
Sales = 2491.7 - 712 Qtr1 - 1512 Qtr2 + 327 Qtr3
c. The quarterly forecasts for next year follow:
Quarter 1 forecast = 2,491.7 – 712(1) – 1,512(0) + 327(0) = 1,779.7 or 1,780
Quarter 2 forecast = 2,491.7 – 712(0) – 1,512(1) + 327(0) = 979.7 or 980
d. The regression output follows.
Regression Equation
Sales = 2306.7 - 642.3 Qtr1 - 1465.4 Qtr2 + 349.8 Qtr3 + 23.13 t
The quarterly forecasts for next year follow:
Quarter 1 forecast = 2,306.7 – 642.3(1) – 1,465.4(0) + 349.8(0) + 23.13(17) =
2,057.61 or 2,058
1,257.64 or 1,258
3,095.97 or 3,096
Quarter 4 forecast = 2,306.7 – 642.3(0) – 1,465.4(0) + 349.8(0) + 23.13(20) = 2,769.3
or 2,769
25.
26.
27.
28.
29. a. Multiplicative Model
Data sales
Length 17
NMissing 0
Fitted Trend Equation
Yt = 204.59 + 0.762*t
Seasonal Indices
Period Index
1 0.96709
2 1.00077
3 1.05435
4 0.97779
Accuracy Measures
MAPE 1.9864
MAD 4.2124
MSD 24.0568
b. Forecasts
Period Forecast
18 218.476
19 230.977
20 214.951
21 213.336
30. See EXCEL Summary below (The forecasts correspond with the last four points of the
graph following.) The modelling undertaken seems both appropriate and successful.
31.
Time Series Decomposition Plot for swimming pool attendances

Multiplicative Model
8 Variable
Actual
Fits
Trend
swimming pool attendances
7
Forecasts
Accuracy Measures
MAPE 5.36430
6
MAD 0.26473
MSD 0.10288
3
2 4 6 8 10 12 14 16 18 20
Index
Time Series Decomposition for swimming pool attendances

Method
Model type Multiplicative Model

Data swimming pool attendances
Length 17
NMissing 1
Fitted Trend Equation

Yt = 3.792 + 0.1507×t
Seasonal Indices
Period Index
1 1.06152
2 1.16595
3 0.93627
4 0.83627
Accuracy Measures
MAPE 5.36430
MAD 0.26473
MSD 0.10288
Forecasts
Period Forecast
18 7.58512
19 6.23205
20 5.69253
21 7.38578
From the graph above the modelling appears to work very well.
Supplementary Exercises:
32. Moving averages often are used to identify movements in stock prices. Daily closing
prices (in dollars per share) for SanDisk for August 16, 2002, through September 3, 2002,
follow (http://www.finance.yahoo.com).
Day Price ($) Day Price ($)

August 16 14.45 August 26 16.45
August 19 15.75 August 27 15.60
August 20 16.45 August 28 15.09
August 21 17.40 August 29 16.42
August 22 17.32 August 30 16.21
August 23 15.96 September 3 15.22
a. Use a five-month moving average to smooth the time series. Forecast the closing price
for September 4, 2002.
b. Use a four-month weighted moving average to smooth the time series. Use a weight of
0.4 for the most recent period, 0.3 for the next period back, 0.2 for the third period
back, and 0.1 for the fourth period back. Forecast the closing price for September 4,
2002.
c. Use exponential smoothing with a smoothing constant of α = 0.7 to smooth the time
series. Forecast the closing price for September 4, 2002.
d. Which of the three methods do you prefer? Why?
33. A chain of grocery stores noted the weekly demand (in cases) reported in the following
table for a particular brand of automatic dishwasher detergent. Use exponential smoothing
with α = 0.2 to develop a forecast for week 11.
Week Demand Week Demand

1 22 6 24
2 18 7 20
3 23 8 19
4 21 9 18
5 17 10 21
34. European Dairies supplies milk to several independent grocers throughout the
Netherlands. Managers at European Dairies want to develop a forecast of the number of
half-litres of milk sold per week. Sales data for the past 12 weeks follow.
Week Sales Week Sales

1 2750 7 3300
2 3100 8 3100
3 3250 9 2950
4 2800 10 3000
5 2900 11 3200
6 3050 12 3150
Use exponential smoothing with α = 0.4 to develop a forecast of demand for week 13.
35. The Garden Avenue Seven sells tapes of its musical performances. The following table
reports sales (in units) for the past 18 months. The group’s manager wants an accurate
method for forecasting future sales.
Month Sales Month Sales Month Sales

1 293 7 381 13 549
2 283 8 431 14 544
3 322 9 424 15 601
4 355 10 433 16 587
5 346 11 470 17 644
6 379 12 481 18 660
a. Use exponential smoothing with α = 0.3, 0.4, and 0.5. Which value of α provides the
best forecasts?
b. Use trend projection to provide a forecast. What is the value of MSE?
c. Which method of forecasting would you recommend to the manager? Why?
36. The Mayer Department Store in Cologne, Germany is trying to determine the amount of
sales lost while it was shut down during July and August because of damage caused by
floods by River Rhine. Sales data for January through June follow.
Month Sales (€1000s) Month Sales (€1000s)
January 185.72 April 210.36
February 167.84 May 255.57
March 205.11 June 261.19
a. Use exponential smoothing, with α = 0.4, to develop a forecast for July and August.
(Hint: Use the forecast for July as the actual sales in July in developing the August
forecast.) Comment on the use of exponential smoothing for forecasts more than one
period into the future.
b. Use trend projection to forecast sales for July and August.
c. Mayer’s insurance company proposed a settlement based on lost sales of €240,000 in
July and August. Is this amount fair? If not, what amount would you recommend as a
counteroffer?
37. Canton Produits is a service firm that employs approximately 100 individuals. Managers
of Canton Produits are concerned about meeting monthly cash obligations and want to
develop a forecast of monthly cash requirements. Because of a recent change in operating
policy, only the past seven months of data are considered to be relevant. With the
following historical data, use trend projection to develop a forecast of cash requirements
for each of the next two months.
Month 1 2 3 4 5 6 7
Cash Required (€1000s) 205 212 218 224 230 240 246
38. The Costello Music Company has been in business for five years. During that time, sales
of electric organs increased from 12 units in the first year to 76 units in the most recent
year. Seamus Costello, the firm’s owner, wants to develop a forecast of organ sales for
the coming year. The historical data follow.
Year 1 2 3 4 5
Sales 12 28 34 50 76
a. Show a graph of this time series. Does a linear trend appear to be present?
b. Develop the equation for the linear trend component for the time series. What is the
average increase in sales that the firm has been realizing per year?
39. Arno Marina has been an authorized dealer for C&D marine radios for the past seven
years. The following table reports the number of radios sold each year.
Year 1 2 3 4 5 6 7
Number Sold 35 50 75 90 105 110 130
a. Show a graph of this time series. Does a linear trend appear to be present?
b. Develop the equation for the linear trend component of the time series.
c. Use the linear trend developed in part (b) to develop a forecast for annual sales in year 8.
40. Refer to the Arno Marina problem in exercise 26. Suppose the quarterly sales values for
the seven years of historical data are as follow.
Total
Year Quarter 1 Quarter 2 Quarter 3 Quarter 4 Yearly Sales
1 6 15 10 4 35
2 10 18 15 7 50
3 14 26 23 12 75
4 19 28 25 18 90
5 22 34 28 21 105
6 24 36 30 20 110
7 28 40 35 27 130
a. Show the four-quarter moving average values for this time series. Plot both the original
time series and the moving average series on the same graph.
b. Compute the seasonal indices for the four quarters.
c. When does Arno Marina experience the largest seasonal effect? Does this result seem
reasonable? Explain.
41. Consider the Costello Music Company problem in exercise 25. The quarterly sales data
follow.
Total
Year Quarter 1 Quarter 2 Quarter 3 Quarter 4 Yearly Sales
1 4 2 1 5 12
2 6 4 4 14 28
3 10 3 5 16 34
4 12 9 7 22 50
5 18 10 13 35 76
a. Compute the seasonal indices for the four quarters.

b. When does Costello Music experience the largest seasonal effect? Does this result
appear reasonable? Explain.
42. Refer to the Arno Marina data in exercise 26.

a. Deseasonalize the data and use the deseasonalized time series to identify the trend.
b. Use the results of part (a) to develop a quarterly forecast for next year based on trend.
c. Use the seasonal indices developed in exercise 27 to adjust the forecasts developed in
part (b) to account for the effect of season.
43. Consider the Costello Music Company time series in exercise 25.
a. Deseasonalize the data and use the deseasonalized time series to identify the trend.
b. Use the results of part (a) to develop a quarterly forecast for next year based on trend.
c. Use the seasonal indices developed in exercise 28 to adjust the forecasts developed in
part (b) to account for the effect of season.
44. The table below shows the number of passengers per quarter (in thousands) who flew with a
charter airline during the years 2003-2005 and the first quarter of 2006.
Year Quarter Passengers Year Quarter Passengers

2003 1 44 2005 1 64
2 92 2 124
3 156 3 200
4 68 4 104
2004 1 60 2006 1 76
2 112
3 180
4 80
a. Derive a multiplicative model for the data and use it to estimate the next three
observations in the series.
b. Graph your results. How would you rate the success of your modelling?
45. A company producing torches is negotiating with a company for the supply of torch bulbs.
The bulb company therefore needs to plan its production to meet the needs of the torch
company and thus uses that company's quarterly sales figures over the past three years to
forecast future demand.
The sales figures are as follows:
Quarterly sales figures (000’s)
Year Ql Q2 Q3 Q4
1 349.4 295.5 196.9 389.3
2 447.5 418 324.1 456.4
3 550.6 528.6 415.2 615.3
a. Using a multiplicative model, estimate trend values and seasonal indices for the series.
Fit a least squares regression line to the trend values. Then use this trend regression
line and your estimates of the seasonal variation factors to forecast future demand for
the four quarters of year 4 for torches.
b. Graph your results and hence comment on the quality of your modelling.
46. Data on road casualties in Great Britain (Wizniewski, 2002) for children (aged under
sixteen) by quarter over four years are as follows:
Year Ql Q2 Q3 Q4
1 8853 12107 13233 10642
2 9338 12138 12527 10543
3 9121 12019 12109 10196
4 8971 11344 12053 9683
a. Decompose the series into trend, seasonal and random components (using a
multiplicative model). Hence derive quarterly forecasts of road casualties for
children in Britain in year 5.
b. By plotting actual and estimated results by quarter make a judgment on the

effectiveness of your modelling.
47. The data below relates to the UK and shows the number of marriages (‘000’s) over a
recent four year period.
Year Quarter Marriages Year Quarter Marriages

1 1 52.9 3 1 41.7
2 114.3 2 100.5
3 138.7 3 138.5
4 62.7 4 60.9
2 1 45.6 4 1 41.7
2 101.9 2 100.5
3 146.2 3 138.5
4 62.3 4 60.9
a. Using the decomposition method, forecast marriages for the next four quarters in the
series.
Supplementary Exercises Solutions:

32. a.
5-Day Moving Forecast
Day Time-Series Average Forecast Error (Error)2
Value
1 14.45
2 15.75
3 16.45
4 17.40
5 17.32
6 15.96 16.27 -0.31 0.10
7 16.45 16.58 -0.13 0.02
8 15.60 16.72 -1.12 1.25
9 15.09 16.55 -1.46 2.12
10 16.42 16.08 0.34 0.11
11 16.21 15.90 0.31 0.09
12 15.22 15.95 -0.73 0.54
Note: MSE = 4.23/7 = 0.60
Forecast for September 4 is (15.60 + 15.09 + 16.42 + 16.21 + 15.22)/5 = 15.71
b. The weighted moving average forecasts for days 5-12 are 16.49, 17.01, 16.71,
16.57, 16.10, 15.60, 15.09, 16.42, 16.21 and 15.22
Note: MSE = 5.21/8 = 0.65
Forecast for September 4 is 0.1(15.09) + 0.2(16.42) + 0.3(16.21) + 0.4(15.22) =

15.74
c. The exponential smoothing forecasts for days 2-12 are 14.45, 15.36, 16.12,
17.02, 17.23, 16.34, 16.42, 15.85, 15.32, 16.09 and 16.17
Note: MSE = 9.57/11 = 0.87
Forecast for September 4 is 0.7(15.22) + 0.3(16.17) = 15.51
d.
Method MSE
Moving Averages 0.60
Weighted Moving Average 0.65
Exponential Smoothing 0.87
Moving Averages is the best of the three approaches because it has the smallest
MSE.
33.
Time-Series Forecast Forecast Error Squared Error

Week t Value Ft Yt - Ft (Yt - Ft)2
Yt
1 22
2 18 22.00 -4.00 16.00
3 23 21.20 1.80 3.24
4 21 21.56 -0.56 0.31
5 17 21.45 -4.45 19.80
6 24 20.56 3.44 11.83
7 20 21.25 -1.25 1.56
8 19 21.00 -2.00 4.00
9 18 20.60 -2.60 6.76
10 21 20.08 0.92 0.85
Total 64.35
MSE = 64.35 / 9 = 7.15
Forecast for week 11:

F11 = 0.2(21) + 0.8(20.08) = 20.26
34.
t Yt Ft Yt - Ft (Yt - Ft)2
1 2,750
2 3,100 2,750.00 350.00 122,500.00
3 3,250 2,890.00 360.00 129,600.00
4 2,800 3,034.00 -234.00 54,756.00
5 2,900 2,940.40 -40.40 1,632.16
6 3,050 2,924.24 125.76 15,815.58
7 3,300 2,974.54 325.46 105,924.21
8 3,100 3,104.73 -4.73 22.37
9 2,950 3,102.84 -152.84 23,260.07
10 3,000 3,041.70 -41.70 1,738.89
11 3,200 3,025.02 174.98 30,618.00
12 3,150 3,095.01 54.99 3,023.90
Total: 488,991.18
MSE = 488,991.18 / 11 = 44,453.74
Forecast for week 13: F13 = 0.4(3,150) + 0.6(3,095.01) = 3,117.01
35. a.
Smoothing Constant MSE
 = 0.3 4,492.37
 = 0.4 2,964.67
 = 0.5 2,160.31
The  = 0.5 smoothing constant is better because it has the smallest MSE.
b. Tt = 244.778 + 22.088t
MSE = 357.81
c. Trend projection provides much better forecasts because it has the
smallest MSE. The reason MSE is smaller for trend projection is that
sales are increasing over time; as a result, exponential smoothing
continuously underestimates the value of sales. If you look at the
forecast errors for exponential smoothing, you will see that the forecast
errors are positive for periods 2 through 18.
36. a. Forecast for July is 236.97
Forecast for August, using forecast for July as the actual sales in July, is
236.97.
Exponential smoothing provides the same forecast for every period in

the future. This is why it is not usually recommended for long-term
forecasting.
b. Tt = 149.719 + 18.451t
Forecast for July is 278.88
Forecast for August is 297.33
c. The proposed settlement is not fair since it does not account for the
upward trend in sales. Based upon trend projection, the settlement
should be based on forecasted lost sales of $278,880 in July and
$297,330 in August.
37. The following values are needed to compute the slope and intercept:
Computation of slope:
Computation of intercept:
225 - 6.8214(4) = 197.714

Equation for linear trend: Tt = 197.714 + 6.821t
Forecast: T8 = 197.714 + 6.821(8) = 252.28
T9= 65.025 + 4.735(9) = 259.10
38. a. A graph of these data shows a linear trend.
b. The following values are needed to compute the slope and intercept:
40 - 15(3) = -5
Equation for linear trend: Tt = -5 + 15t
Conclusion: average increase in sales is 15 units per year
39. a. Yes, a linear trend appears to exist.
b. The following values are needed to compute the slope and intercept:
85 - 15.5357(4) = 22.857
Equation for linear trend: Tt = 22.857 + 15.536t
c. Forecast: T8 = 22.857 + 15.536(8) = 147.15
40. a.
Centred Seasonal-
t Sales Moving Average Irregular
Component
1 6
2 15
3 10 9.250 1.081
4 4 10.125 0.395
5 10 11.125 0.899
6 18 12.125 1.485
7 15 13.000 1.154
8 7 14.500 0.483
9 14 16.500 0.848
10 26 18.125 1.434
11 23 19.375 1.187
12 12 20.250 0.593
13 19 20.750 0.916
14 28 21.750 1.287
15 25 22.875 1.093
16 18 24.000 0.750
17 22 25.125 0.876
18 34 25.875 1.314
19 28 26.500 1.057
20 21 27.000 0.778
21 24 27.500 0.873
22 36 27.625 1.303
23 30 28.000 1.071
24 20 29.000 0.690
25 28 30.125 0.929
26 40 31.625 1.265
27 35
28 27
b.
Seasonal-Irregular Seasonal
Quarter Component Values Index
1 0.899, 0.848, 0.916, 0.876, 0.873, 0.929 0.890
2 1.485, 1.434, 1.287, 1.314, 1.303, 1.265 1.348
3 1.081, 1.154, 1.187, 1.093, 1.057, 1.071 1.107
4 0.395, 0.483, 0.593, 0.750, 0.778, 0.690 0.615
Total 3.960
Quarter Adjusted Seasonal Index

1 0.899
2 1.362
3 1.118
4 0.621
Note: Adjustment for seasonal index = 4.00 / 3.96 = 1.0101
c. Hudson Marine experiences the largest seasonal increase in quarter 2. Since

this quarter occurs prior to the peak summer boating season, this result seems
reasonable.
41. a.
Centred Seasonal-Irregular
t Sales Moving Average Component
1 4
2 2
3 1 3.250 0.308
4 5 3.750 1.333
5 6 4.375 1.371
6 4 5.875 0.681
7 4 7.500 0.533
8 14 7.875 1.778
9 10 7.875 1.270
10 3 8.250 0.364
11 5 8.750 0.571
12 16 9.750 1.641
13 12 10.750 1.116
14 9 11.750 0.766
15 7 13.250 0.528
16 22 14.125 1.558
17 18 15.000 1.200
18 10 17.375 0.576
19 13
20 35
Seasonal-Irregular Seasonal
Quarter Component Values Index
1 1.371, 1.270, 1.116, 1.200 1.239

2 0.681, 0.364, 0.776, 0.576 0.597
3 0.308, 0.533, 0.571, 0.528 0.485
4 1.333, 1.778, 1.641, 1.558 1.578
Total 3.899
Quarter Adjusted Seasonal Index

1 1.271
2 0.613
3 0.498
4 1.619
Note: Adjustment for seasonal index = 4 / 3.899 = 1.026
b. The largest effect is in quarter 4; this seems reasonable since retail sales are
generally higher during October, November, and December.
42. a. Note: To simplify the calculations the seasonal indexes calculated in problem
27 have been rounded to two decimal places.
Seasonal Factor Deseasonalized Sales

Year Quarter Sales Yt St Yt / St = TtIt
1 1 6 0.90 6.67
2 15 1.36 11.03
3 10 1.12 8.93
4 4 0.62 6.45
2 1 10 0.90 11.11
2 18 1.36 13.24
3 15 1.12 13.39
4 7 0.62 11.29
3 1 14 0.90 15.56
2 26 1.36 19.12
3 23 1.12 20.54
4 12 0.62 19.35
4 1 19 0.90 21.11
2 28 1.36 20.59
3 25 1.12 22.32
4 18 0.62 29.03
5 1 22 0.90 24.44
2 34 1.36 25.00
3 28 1.12 25.00
4 21 0.62 33.87
6 1 24 0.90 26.67
2 36 1.36 26.47
3 30 1.12 26.79
4 20 0.62 32.26
7 1 28 0.90 31.11
2 40 1.36 29.41
3 35 1.12 31.25
4 27 0.62 43.55
Yt
t (deseasonalized) tYt t2
1 6.67 6.67 1
2 11.03 22.06 4
3 8.93 26.79 9
4 6.45 25.80 16
5 11.11 55.55 25
6 13.24 79.44 36
7 13.39 93.73 49
8 11.29 90.32 64
9 15.56 140.04 81
10 19.12 191.20 100
11 20.54 225.94 121
12 19.35 232.20 144
13 21.11 274.43 169
14 20.59 288.26 196
15 22.32 334.80 225
16 29.03 464.48 256
17 24.44 415.48 289
18 25.00 450.00 324
19 25.00 475.00 361
20 33.87 677.40 400
21 26.67 560.07 441
22 26.47 582.34 484
23 26.79 616.17 529
24 32.26 774.24 576
25 31.11 777.75 625
26 29.41 764.66 676
27 31.25 843.75 729
28 43.55 1,219.40 784
406 605.55 10,707.34 7,714
b/c.
t Trend Forecast
29 36.92
30 37.98
31 39.03
32 40.09
Year Quarter Trend Forecast Seasonal Index Quarterly

Forecast
8 1 36.92 0.90 33.23
2 37.98 1.36 51.65
3 29.03 1.12 43.71
4 40.09 0.62 24.86
43. a Note: To simplify the calculations the seasonal indexes in problem 28 have
been rounded to two decimal places.
Seasonal Factor Deseasonalized Sales
Year Quarter Sales Yt St Yt / St = TtIt
1 1 4 1.27 3.15
2 2 0.61 3.28
3 1 0.50 2.00
4 5 1.62 3.09
2 1 6 1.27 4.72
2 4 0.61 6.56
3 4 0.50 8.00
4 14 1.62 8.64
3 1 10 1.27 7.87
2 3 0.61 4.92
3 5 0.50 10.00
4 16 1.62 9.88
4 1 12 1.27 9.45
2 9 0.61 14.75
3 7 0.50 14.00
4 22 1.62 13.58
5 1 18 1.27 14.17
2 10 0.61 16.39
3 13 0.50 26.00
4 35 1.62 21.60
Yt
t2
t (deseasonalized) tYt
1 3.15 3.15 1
2 3.28 6.56 4
3 2.00 6.00 9
4 3.09 12.36 16
5 4.72 23.60 25
6 6.56 39.36 36
7 8.00 56.00 49
8 8.64 69.12 64
9 7.87 70.83 81
10 4.92 49.20 100
11 10.00 110.00 121
12 9.88 118.56 144
13 9.45 122.85 169
14 14.75 206.50 196
15 14.00 210.00 225
16 13.58 217.28 256
17 14.17 240.89 289
18 16.39 295.02 324
19 26.00 494.00 361
20 21.60 432.00 400
210 202.05 2783.28 2870
b.
y Trend Forecast
21 20.55
22 21.55
23 22.54
24 23.54
c.
Trend Seasonal Quarterly
Year Quarter Forecast Index Forecast
6 1 20.55 1.27 26.10
2 21.55 0.61 13.15
3 22.54 0.50 11.27
4 23.54 1.62 38.13
44. From EXCEL we have

Centred Seasonal Seasonal
Passengers Moving Irregular Factor Deseasonalised
Year Quarter Yt Average Component St Passengers
2003 1 44 0.5759 76.40
2 92 1.0466 87.90
3 156 92.00 1.6957 1.6603 93.96
4 68 96.50 0.7047 0.7172 94.81
2004 1 60 102.00 0.5882 0.5759 104.18
2 112 106.50 1.0516 1.0466 107.01
3 180 108.50 1.6590 1.6603 108.41
4 80 110.50 0.7240 0.7172 111.54
2005 1 64 114.50 0.5590 0.5759 111.13
2 124 120.00 1.0333 1.0466 118.48
3 200 124.50 1.6064 1.6603 120.46
4 104 0.7172 145.01
2006 1 76 0.5759 131.97
where
Quarter
1 2 3 4
2003 1.6957 0.7047
200
Year 4 0.5882 1.0516 1.6590 0.7240
2005 0.5590 1.0333 1.6064
Seasonal Index 0.5736 1.0425 1.6537 0.7143 3.9841
Adjusted Seasonal Index 0.5759 1.0466 1.6603 0.7172 4.0000
=7 = 104.6154 b1 = 4.4744 b0 = 77.239 Tt = 77.239 + 4.4744t
y Trend Forecast
14 139.88
15 144.36
16 148.83

2006 2 139.88 1.0466 146
3 144.36 1.6603 240
4 148.83 0.7172 107
45. a. From EXCEL we have

Sales Moving Irregular Factor Deseasonalised
Year Quarter Yt Average Compone St Sales
nt
1 1 349.4 1.1795 296.23
2 295.5 1.0489 281.72
3 196.9 320.04 0.6152 0.6959 282.94
4 389.3 347.61 1.1199 1.0758 361.87
2 1 447.5 378.83 1.1813 1.1795 379.40
2 418 403.11 1.0369 1.0489 398.51
3 324.1 424.39 0.7637 0.6959 465.73
4 456.4 451.10 1.0117 1.0758 424.24
3 1 550.6 476.31 1.1560 1.1795 466.81
2 528.6 507.56 1.0414 1.0489 503.96
3 415.2 0.6959 596.64
4 615.3 1.0758 571.95
where Quarter
1 2 3 4
1 0.6152 1.1199
Year 2 1.1813 1.0369 0.7637 1.0117
3 1.1560 1.0414
Seasonal Index 1.1686 1.0392 0.6895 1.0658 3.9631
= 6.5 = 419.17 b1 = 28.464 b0 = 234.15 Tt = 234.15 + 28.464t
y Trend Forecast
13 604.18
14 632.65
15 661.11
16 689.57

4 1 604.18 1.1795 712.63
2 632.65 1.0489 663.58
3 661.11 0.6959 460.07
4 689.57 1.0758 741.84
b. The model seems to be of high quality from the plot below.

800
700
600
500
400
300
200
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
46. a. From EXCEL we have

Casualties Moving Irregular Factor Deseasonalised
Year Quarter Yt Average Component St Casualties
1 1 8853 0.8322 10638.07
2 12107 1.0856 11152.36
3 13233 11269.38 1.1742 1.1372 11636.48
4 10642 11333.88 0.9390 0.945 11261.38
2 1 9338 11249.50 0.8301 0.8322 11220.86
2 12138 11148.88 1.0887 1.0856 11180.91
3 12527 11109.38 1.1276 1.1372 11015.65
4 10543 11067.38 0.9526 0.945 11156.61
3 1 9121 11000.25 0.8292 0.8322 10960.11
2 12019 10904.63 1.1022 1.0856 11071.30
3 12109 10842.50 1.1168 1.1372 10648.08
4 10196 10739.38 0.9494 0.945 10789.42
4 1 8971 10648.00 0.8425 0.8322 10779.86
2 11344 10576.88 1.0725 1.0856 10449.52
3 12053 1.1372 10598.84
4 9683 0.945 10246.56
where
Quarter
1 2 3 4
1 1.1742 0.9390
Year 2 0.8301 1.0887 1.1276 0.9526
3 0.8292 1.1022 1.1168 0.9494
4 0.8425 1.0725
Seasonal Index 0.8339 1.0878 1.1396 0.9470 4.0083
= 8.5 = 10925.38 b1 = -53.19 b0 = 11378 Tt = 11378 - 53.195t
y Trend Forecast
17 10473.69
18 10420.49
19 10367.30
20 10314.10

5 1 10473.69 0.8322 8716
2 10420.49 1.0856 11312
3 10367.30 1.1372 11790
4 10314.10 0.945 9747
b. From the EXCEL plot below the model looks very effective.
14000
12000
10000
8000
6000
4000
2000
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
47. a.

ch17 Solutions

Uploaded by

Copyright:

Available Formats

ch17 Solutions

Uploaded by

Document Information

Original Title

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

ch17 Solutions

Uploaded by

Copyright:

Available Formats

Chapter 17: Forecasting

4. Consider the following time series data.

5. Consider the following time series data.

6. Consider the following time series data.

2013 Jun 1,177 2016 Mar 945

14. Consider the following time series data.

16. Consider the following time series.

a. Construct a time series plot. What type of pattern

17. Consider the following time series.

Year Period Users (billions)

a. Construct a time series plot. What type of pattern exists?

2009 2010 2011 2012 2013 2014 2015 2016 2017

GDP (USD billion) for 2008–17 are tabulated below (tradingeconomics.com/singapore/gdp):

Year USD (billions)

a. Graph this time series. Does a linear trend appear to be present?

22 Consider the following time series.

Quarter Year 1 Year 2 Year 3

23 Consider the following time series data.

Quarter Year 1 Year 2 Year 3

Quarter Year 1 Year 2 Year 3

26 Consider the following time series data.

Quarter Year 1 Year 2 Year 3

27 Refer to exercise 26.

Quarter Year 1 Year 2 Year 3

Comment on the quality of your modelling.

Textbook Exercises Solutions:

MSE = 363/6 = 60.5

The data appear to follow a horizontal pattern.

MSE = 65.15/5 = 13.03

MSE = 110/4 = 27.5.

b. Smoothing constant = 0.2

The following tables show the calculations for α = 0.2.

Moving Average Plot for C1

Smoothing Plot for C1

Period Forecast Lower Upper

The data appear to follow a horizontal pattern.

13. a. MA3 results

Moving Average Plot for NZ merchandise imports

Based on diagnostics (accuracy measures) provided, the 3-point moving average is

The time-series plot exhibits a curvilinear trend.

A very faintly non-linear pattern e.g. quadratic in evidence here.

Trend Analysis Plot for Social media users

Quadratic trend results above represent almost a perfect fit.

Trend Analysis Plot for Visitors to Ireland

b. Nevertheless, from EXCEL

Equation for linear trend: Tt = 5301 + 376.4xt

Hence forecasts are as follows:

T2018 = 9062 when xt = 10

T2019 = 9438 when xt = 11

T2020 = 9814 when xt = 12

Trend Analysis Plot for Singapore GDP

Nevertheless, from EXCEL

Equation for linear trend: Tt =192.0 + 14.94xt

c. Hence forecasts are as follows:

T2018 = 356 when xt = 11

T2019 = 371 when xt = 12

T2020 = 386 when xt = 13

b. A portion of the Minitab regression output is shown;

1 1.3333,1.0980 1.2157 1.2050