13-1 of 26

Chapter Thirteen
Time Series Forecasting

McGraw-Hill/Irwin Copyright © 2003 by The McGraw-Hill Companies, Inc. All rights reserved.
13-2 of 26
 Time Series Forecasting
13.1 Time Series Components and Models
13.2 Time Series Regression: Basic Models
13.3* Time Series Regression: More Advanced
Models
13.4 Multiplicative Decomposition
13.5 Exponential Smoothing
13.6 Forecast Error Comparisons
13.7 Index Numbers

13-3 of 26
 13.1 Time Series Components and
Models

Trend Long-run growth or decline.

Cycle Long-run up and down fluctuation around the trend level.
Seasonal Regular periodic up and down movements that repeat within
the calendar year.
Irregular Erratic very short-run movements that follow no regular
pattern.

13-4 of 26
 13.2 Time Series Regression: No
Trend
Example 13.1: The Cod Catch Case
Cod Catch (in Tons)

Month Year 1 Year 2

Jan 362 276
Feb 381 334
Mar 317 394
Apr 297 334
May 399 384
Jun 402 314
Jul 375 344
Aug 349 337
Sep 386 345
Oct 328 362
Nov 389 314
Dec 343 365 “No Trend” Regression Model
yt = β0  ε t
yˆ t b0 351.29
13-5 of 26
 Time Series Regression: Linear Trend

Example 13.2: The Calculator Sales Case

Ca lcu la to r S a le s Da ta

M o n th Ye a r 1 Ye a r 2
Jan 197 296
F eb 211 276
M ar 203 305
A pr 247 308
M ay 239 356
Jun 269 393
Jul 308 363
A ug 262 386
S ep 258 443
Oct 256 308
Nov 261 358
Dec 288 384
yˆ t b0  b1t 198.03  8.0743(t )
Linear Trend Regression yˆ 25 198.03  8.0743(25) 399.9
yt = β0  1t  εt yˆ 26 198.03  8.0743(26) 408.0

13-6 of 26
 Time Series Regression: Seasonal
Regression
Example 13.3: The Bike Sales Case

Year Qtr t Sales, yt

1 1 1 10
2 2 31
3 3 43
4 4 16
2 1 5 11
2 6 33
3 7 45
4 8 17
3 1 9 13
2 10 34
3 11 48
4 12 19
4 1 13 15 Seasonal Regression
2 14 37
3 15 51 yt = β0  1t   Q 2Q2   Q 3Q3   Q 4Q4  εt
4 16 21

13-7 of 26
 Example: Seasonal Regression

Example 13.3: The Bike Sales Case Minitab Output

BikeSales = 8.75 + 0.500 Time + 21.0 Q2 + 33.5 Q3 + 4.50 Q4
Predictor Coef StDev T P
Constant 8.7500 0.4281 20.44 0.000
Time 0.50000 0.03769 13.27 0.000
Q2 21.0000 0.4782 43.91 0.000
Q3 33.5000 0.4827 69.41 0.000
Q4 4.5000 0.4900 9.18 0.000
S = 0.6742 R-Sq = 99.8% R-Sq(adj) = 99.8%
Predicted Values (Time = 17, 18, 19, 20)
Fit StDev Fit 95.0% CI 95.0% PI
17.250 0.506 ( 16.137, 18.363) ( 15.395, 19.105)
38.750 0.506 ( 37.637, 39.863) ( 36.895, 40.605)
51.750 0.506 ( 50.637, 52.863) ( 49.895, 53.605)
23.250 0.506 ( 22.137, 24.363) ( 21.395, 25.105)
yˆ17 8.75  0.5(17)  21(0)  33.5(0)  4.5(0) 17.25
...
yˆ 20 8.75  0.5(20)  21(0)  33.5(0)  4.5(1) 23.25

13-8 of 26
 13.3* Time Series Regression:
Advanced
Example 13.4: Traveler’s Rest Case

Seasonal Regression
yt.25 = β0  1t   M 1M 1   M 2 M 2  ...   M 11 M 11  εt

13-9 of 26
 Example: Advanced Seasonal
Regression
Example 13.4: Traveler’s Rest Case MegaStat Output
Regression output confidence interval
variables coefficients std. error t (df=155) p-value 95% lower 95% upper
Intercept 4.807318 0.00846255 568.070 4.06E-259 4.7906 4.8240
t 0.003515 0.00004449 79.009 3.95E-127 0.0034 0.0036
M1 -0.052467 0.01055475 -4.971 1.75E-06 -0.0733 -0.0316
M2 -0.140790 0.01055278 -13.342 1.59E-27 -0.1616 -0.1199
M3 -0.107103 0.01055100 -10.151 7.02E-19 -0.1279 -0.0863
M4 0.049882 0.01054940 4.728 5.05E-06 0.0290 0.0707
M5 0.025417 0.01054800 2.410 .0171 0.0046 0.0463
M6 0.190170 0.01054678 18.031 6.85E-40 0.1693 0.2110
M7 0.382455 0.01054575 36.266 1.28E-77 0.3616 0.4033
M8 0.413370 0.01054490 39.201 2.41E-82 0.3925 0.4342
M9 0.071417 0.01054424 6.773 2.47E-10 0.0506 0.0922
M10 0.050641 0.01054377 4.803 3.66E-06 0.0298 0.0715
M11 -0.141943 0.01054349 -13.463 7.47E-28 -0.1628 -0.1211
Durbin-Watson = 1.26
13-10 of 26
 Example: Advanced Seasonal
Regression (Cont.)
Example 13.4: Traveler’s Rest Case MegaStat Output
Predicted values for: TFY2
95% Confidence Intervals 95% Prediction Intervals
t Predicted lower upper lower upper Leverage
169 5.3489 5.3322 5.3656 5.2913 5.4065 0.092
170 5.2641 5.2474 5.2808 5.2065 5.3217 0.092
171 5.3013 5.2846 5.3180 5.2437 5.3589 0.092
172 5.4618 5.4451 5.4785 5.4042 5.5194 0.092
173 5.4409 5.4241 5.4576 5.3833 5.4984 0.092
174 5.6091 5.5924 5.6258 5.5515 5.6667 0.092
175 5.8049 5.7882 5.8216 5.7473 5.8625 0.092
176 5.8394 5.8226 5.8561 5.7818 5.8969 0.092
177 5.5009 5.4842 5.5176 5.4433 5.5585 0.092
178 5.4837 5.4669 5.5004 5.4261 5.5412 0.092
179 5.2946 5.2779 5.3113 5.2370 5.3522 0.092
180 5.4400 5.4233 5.4568 5.3825 5.4976 0.092
.25
yˆ169 4.807318  0.003515(169)  (  0.0524467)(1) 5.3489
yˆ169  yˆ169  5.3489  818.57;
4
.25 4
95% P. I. [(5.2913)4 , (5.4065)4 ] [783.91, 854.36]
13-11 of 26
 13.4 Multiplicative Decomposition
Example 13.5: The Tasty Cola Case

Year Mo t Sales, yt Year Mo t Sales, yt

1 1 1 189 2 7 19 831
2 2 229 8 20 960
3 3 249 9 21 1152
4 4 289 10 22 759
5 5 260 11 23 607
6 6 431 12 24 371
7 7 660 3 1 25 298
8 8 777 2 26 378
9 9 915 3 27 373
10 10 613 4 28 443
11 11 485 5 29 374
12 12 277 6 30 660
2 1 13 244 7 31 1004
2 14 296 8 32 1153
3 15 319 9 33 1388
4 16 370 10 34 904 Multiplicative Model
5 17 313 11 35 715
6 18 556 12 36 441 yt TRt SN t CLt IRt

13-12 of 26
 Example: Multiplicative
Decomposition
Example 13.5: The Tasty Cola Case

Prediction
Multiplicative Model
t
37
trt
731.273
snt
0.493
trt x snt
360.52
yt
352
yˆ t trt snt
38 740.762 0.596 441.49 445
39 750.252 0.595 446.40 453
40 759.741 0.680 516.62 541
41 769.231 0.564 433.85 457
42 778.720 0.986 767.82 762
43 788.209 1.467 1156.30 1194
44 797.699 1.693 1350.50 1361
45 807.188 1.990 1606.30 1615
46 816.678 1.307 1067.40 1059
47 826.167 1.029 850.13 824
48 835.657 0.600 501.39 495

trt 380.163  9489t

13-13 of 26
 13.5 Exponential Smoothing

Example 13.6: The Cod Catch Case

Cod Catch (in Tons)

Month Year 1 Year 2

Jan 362 276
Feb 381 334
Mar 317 394
Apr 297 334
May 399 384
Jun 402 314
Jul 375 344
Aug 349 337
Sep 386 345
Oct 328 362
Nov 389 314
Dec 343 365
“No Trend” Model
yt = β0  ε t

13-14 of 26
 Simple Exponential Smoothing

Model:
yt = β0  ε t
Smoothing Equation:
ST yt  (1   ) ST  1
 is a smoothing constant, 0 α 1
ST  1 is an estimate of  0 from time T-1
Forecast:
yˆT  = ST for all future values,  1, 2, ...

Updating: (Given yT+1 in time period T+1)

ST 1 yt 1  (1   ) ST
yˆT 1 ST 1 for all future values, 1, 2,...
13-15 of 26
 Example: Simple Exponential
Smoothing
Example 13.7: The Cod Catch Case

 0.07

yˆ 25 S 24 348.168

13-16 of 26
Example: Holt-Winters’ Double Expo
Smoothing
Example: The Calculator Sales Case

Model
yt = β0  β1t  ε

Method

Initial estimates of
β0 , β1
Updating equations

13-17 of 26
 Example: Multiplicative Winters’
Method
Example: The Tasty Cola Case

Model
yt = ( β0  β1t ) SN t  ε

Method

Initial estimates of
β0 , β1 ,
seasonal factors
Updating equations

13-18 of 26
 13.6 Forecast Error Comparison

Forecast Errors et  yt  yˆ t

Error Comparison Criteria

Mean Absolute Deviation (MAD)
n n

e t y t  yˆ t
MAD  t 1  t 1
n n
Mean Squared Deviation (MSD)
n n

e 2
t  t t
( y  ˆ
y ) 2

MSD  t 1  t 1
n n
13-19 of 26
 Example: Forecast Error Comparison
Example: The Tasty Cola Case

Decomposition Winters
2
Time Actual PredD ErrD |ErrD | (ErrD ) PredW ErrW |ErrW | (ErrW )2
37 352 360.52 -8.52 8.52 72.59 355.96 -3.96 3.96 15.68
38 445 441.48 3.52 3.52 12.39 426.31 18.69 18.69 349.32
39 453 446.40 6.60 6.60 43.56 436.69 16.31 16.31 266.02
40 541 516.62 24.38 24.38 594.38 505.60 35.40 35.40 1253.16
41 457 433.85 23.15 23.15 535.92 431.71 25.29 25.29 639.58
42 762 767.82 -5.82 5.82 33.87 748.35 13.65 13.65 186.32
43 1194 1156.30 37.70 37.70 1421.29 1132.57 61.43 61.43 3773.64
44 1361 1350.50 10.50 10.50 110.25 1313.74 47.26 47.26 2233.51
45 1615 1606.30 8.70 8.70 75.69 1576.09 38.91 38.91 1513.99
46 1059 1067.40 -8.40 8.40 70.56 1043.59 15.41 15.41 237.47
47 824 850.12 -26.12 26.12 682.25 834.24 -10.24 10.24 104.86
48 495 501.39 -6.39 6.39 40.83 506.65 -11.65 11.65 135.72

Sum 169.80 3693.60 298.20 10709.27

Mean 14.15 307.80 24.85 892.44
MADD MSED MADw MSEw

13-20 of 26
 13.7 Index Numbers

yt
Simple (Quantity) Index it  100
y0

Example Installment Credit Outstanding (in billions of dollars): 1990 -1996

1990 1991 1992 1993 1994 1995 1996

Installment credit outstanding 796.4 781.1 784.9 844.1 966.5 1103.3 1194.6
Base (base year = 1990) 796.4 796.4 796.4 796.4 796.4 796.4 796.4
Index (current outstanding/base) 100.00 98.08 98.56 105.99 121.36 138.54 150.00

13-21 of 26
 Aggregate Price Index

Aggregate Price Index it 

 p t
100
p 0

Example: 1992 (base) and 1997 Prices for a

Market Basket of Grocery Items
1992 1997
Grocery Item
Bag of Apples
Price
2.99
Price
3.69 i1997 
 p 1997
100
Loaf of Bread 0.99 1.29 p 1992
Tuna Fish 0.69 0.79
10.74
Gallon of Milk 1.29 1.59  100 125.76
Peanut Butter 1.99 2.59 8.54
Green Beans 0.59 0.79
Totals 8.54 10.74

13-22 of 26
 Weighted Aggregate Price Index
(Laspeyres)

Laspeyres Index it 
 pqt 0
100
(base quantity weights) pq 0 0

Example: 1992 (base) and 1997 Prices for a Market Basket of Grocery Items
1992 (Base Year) 1997
1992 1992 1997 1992
Grocery Item Price Quantity pxq Price Quantity pxq
Bag of Apples 2.99 26 77.74 3.69 26 95.94
Loaf of Bread 0.99 156 154.44 1.29 156 201.24
Tuna Fish 0.69 52 35.88 0.79 52 41.08
Gallon of Milk 1.29 104 134.16 1.59 104 165.36
Peanut Butter 1.99 13 25.87 2.59 13 33.67
Green Beans 0.59 52 30.68 0.79 52 41.08
Totals 458.77 578.37

i1997 
p q
1997 1992
100 
578.37
100 126.07
p q
1992 1992 458.77

13-23 of 26
 Weighted Aggregate Price Index
(Paasche)

Paasche Index it 
 pqt t
100
(current quantity weights) pq0 0

Example: 1992 (base) and 1997 Prices for a Market Basket of Grocery Items
1992 (Base Year) 1997
1992 1992 1997 1997
Grocery Item Price Quantity pxq Price Quantity pxq
Bag of Apples 2.99 26 77.74 3.69 52 191.88
Loaf of Bread 0.99 156 154.44 1.29 156 201.24
Tuna Fish 0.69 52 35.88 0.79 104 82.16
Gallon of Milk 1.29 104 134.16 1.59 78 124.02
Peanut Butter 1.99 13 25.87 2.59 8 20.72
Green Beans 0.59 52 30.68 0.79 156 123.24
Totals 458.77 743.26

i1997 
p q
1997 1997
100 
743.26
100 162.01
p q
1992 1992 458.77

13-24 of 26
Example: Laspeyres Price Index,
CPI-U
CPI-U Consumer Price Index for all Urban Workers (1980 – 1996)

13-25 of 26
 Time Series Forecasting

13.1 Time Series Components and Models

Summary:
13.2 Time Series Regression: Basic Models
13.3* Time Series Regression: More Advanced
Models
13.4 Multiplicative Decomposition
13.5 Exponential Smoothing
13.6 Forecast Error Comparisons
13.7 Index Numbers

13-26 of 26