Table 7.

1 Tahoe Salt

Table 7.1 Quarterly Demand for Tahoe Salt

Period Demand
Year, Qtr t Dt
2011,2 1 8,000
2011,3 2 13,000
2011,4 3 23,000
2012,1 4 34,000
2012,2 5 10,000
2012,3 6 18,000
2012,4 7 23,000
2013,1 8 38,000
2013,2 9 12,000
2013,3 10 13,000
2013,4 11 32,000 Forecast Demand
2014,1 12 41,000
2014,2 13 ?
2014,3 14 ?
2014,4 15 ?
2015,1 16 ?

Figure 7.1 Quarterly Demand at Tahoe Salt

45,000
Demand

40,000
38,000
35,000 34,000
32
30,000
25,000
23,000 23,000
20,000
18,000
15,000
13,000 12,000 13,000
10,000 10,000
8,000
5,000
0
2011,2 2011,3 2011,4 2012,1 2012,2 2012,3 2012,4 2013,1 2013,2 2013,3 2013,4

Year, Quarter
 Table 7.1 Tahoe Salt

Tahoe Salt

41,000
38,000

32,000

12,000 13,000

013,1 2013,2 2013,3 2013,4 2014,1

 Figure7.2

Period Demand Deseasonalized

t Dt Demand
1 8,000
2 13,000
3 23,000 19,750
4 34,000 20,625
5 10,000 21,250
6 18,000 21,750
7 23,000 22,500
8 38,000 22,125
9 12,000 22,625
10 13,000 24,125
11 32,000
12 41,000

Equation 7.2 Deseasonalizing Demand

where:
D = Demand
p = periodicity
t = period

Figure 7.3 Deseasonalized Demand for Tahoe Sa

45,000

40,000
Demand

35,000

30,000

25,000

20,000

15,000

10,000

5,000

0
0 1 2 3 4 5 6 7 8 9 10
Period, t
15,000

Figure7.2
10,000

5,000

0
0 1 2 3 4 5 6 7 8 9 10
Period, t
 Figure7.2

and for Tahoe Salt

7 8 9 10 11 12 13
 Figure7.2

7 8 9 10 11 12 13
 Regression 7.2

REGRESSION SUMMARY OUTPUT

Regression Statistics
Multiple R 0.958065237
R Square 0.917888998
Adjusted R Sq 0.90420383
Standard Error 414.5033124
Observations 8

ANOVA
df SS MS F Significance F
Regression 1 11523809.524 11523810 67.07182 0.0001786086
Residual 6 1030877.9762 171813
Total 7 12554687.5

Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 18,439 440.80870788 41.829909 1.249E-08 17360.367255 19517.609 17360.3673 19517.60894
X Variable 1 524 63.959249681 8.1897384 0.000179 367.30676332 680.31228 367.306763 680.3122843

Initial Level, L
Trend, T
 Figure7.4

Deseasonalized Seasonal
Period Demand Demand D t Factor Estimate S i
t Dt (Eqn 7.4) (Eqn 7.5) S (Eqn 7.6)
t Forecast
1 8,000 18,963 0.42 0.47 8,944
2 13,000 19,487 0.67 0.68 13,317
3 23,000 20,011 1.15 1.17 23,426
4 34,000 20,535 1.66 1.66 34,177
5 10,000 21,059 0.47 9,933
6 18,000 21,583 0.83 14,749
7 23,000 22,107 1.04 25,879
8 38,000 22,631 1.68 37,665
9 12,000 23,155 0.52 10,921
10 13,000 23,679 0.55 16,182
11 32,000 24,203 1.32 28,333
12 41,000 24,727 1.66 41,153
13 11,910
Ft + 1

14 17,614
Ft + 1

15 30,787
Ft + 1

16 44,642
Ft + 1

Forecasted Data
Forecasted
Period Demand
Year, Qtr t Ft+1
12,2 13 11,910
12,3 14 17,614
12,4 15 30,787
13,1 16 44,642
 Ex.7-1

Page 225 Example 7-1 Moving Average

Period Demand Level Forecast Error

t Dt Lt Ft Et
1 120
2 127
3 114
4 122 120.75
5 125 122.00 120.75 -4.25
6 122.00
7
8
9
10
11
12
 Ex.7-2

Page 226 Example 7-2 Simple Exponential Smoothing

Demand Forecast
Period t Dt Level Lt Ft Error Et
0 120.75
1 120 120.68 120.75 0.75
2 127 121.31 120.68 -6.33
3 114 120.58 121.31 7.31
4 122 120.72 120.58 -1.42
5 108.65 120.72
6
7
8
9
10
11
12

alpha 0.1
 Ex.7-3

Page 228 Example 7-3 Holt's Model

Demand Forecast
Period t Dt Level Lt Trend Tt Ft Error Et
0 7,367 673
1 8,415 8,078 681 8,040 -375
2 8,732 8,755 680 8,758 26
3 9,014 9,393 672 9,435 421
4 9,808 10,039 666 10,065 257
5 10,413 10,676 661 10,706 293
6 11,961 11,399 673 11,337 -624
7 12,072
8
9
10
11
12

alpha 0.1
Beta 0.2
摘要輸出

迴歸統計
R 的倍數 0.955665
R 平方 0.913295
調整的 R 平方
0.891619
標準誤 433.9514
觀察值個數 6

ANOVA
自由度 SS MS F 顯著值
迴歸 1 7934336 7934336 42.13358 0.002905
殘差 4 753255.3 188313.8
總和 5 8687591

係數 標準誤 t 統計 P-值 下限 95% 上限 95% 下限 95.0%上限 95.0%

截距 7367.133 403.9868 18.23608 5.32E-05 6245.486 8488.78 6245.486 8488.78
X 變數 1 673.3429 103.7342 6.491039 0.002905 385.3305 961.3552 385.3305 961.3552
 Ex.7-4

Page 229 Example 7-4 Winter's Model

Demand Seasonal Factor Forecast

Period t Dt Level Lt Trend Tt St Ft Error Et
0 18,439 524
1 8,000 18,769 485 0.47 8,913 913
2 13,000 19,240 482 0.68 13,093 93
3 23,000 19,716 481 1.17 23,076 76
4 34,000 20,214 484 1.67 33,730 -270
5 10,000 20,776 500 0.47 9,637 -363
6 18,000 21,797 604 0.68 14,458 -3,542
7 23,000 22,127 549 1.17 26,202 3,202
8 38,000 22,683 551 1.67 37,898 -102
9 12,000 23,479 600 0.47 10,855 -1,145
10 13,000 23,543 493 0.69 16,715 3,715
11 32,000 24,399 565 1.16 27,801 -4,199
12 41,000 24,921 557 1.67 41,731 731
13 0.47 12,015
14 0.68 17,703
15 1.17 31,167
16 1.67 45,307

alpha 0.1 L
Beta 0.2 T
Gamma 0.1 S

50,000

45,000

40,000

35,000

30,000
Demand
25,000 Forecast
20,000

15,000
 Figure7.5 Moving Average

Tahoe Salt Forecasts Using Four-Period Moving Average

Absolute Mean Squared
Period Demand Level Forecast Error Error Error
t Dt Lt Ft Et At MSEt MADt % Error MAPEt TSt
1 8,000
2 13,000
3 23,000
4 34,000 19,500
5 10,000 20,000 19,500 9,500 9,500 90,250,000 9,500 95 95 1.00
6 18,000 21,250 20,000 2,000 2,000 47,125,000 5,750 11 53 2.00
7 23,000 21,250 21,250 -1,750 1,750 32,437,500 4,417 8 38 2.21
8 38,000 22,250 21,250 -16,750 16,750 94,468,750 7,500 44 39 -0.93
9 12,000 22,750 22,250 10,250 10,250 96,587,500 8,050 85 49 0.40
10 13,000 21,500 22,750 9,750 9,750 96,333,333 8,333 75 53 1.56
11 32,000 23,750 21,500 -10,500 10,500 98,321,429 8,643 33 50 0.29
12 41,000 24,500 23,750 -17,250 17,250 123,226,563 9,719 42 49 -1.52
13 24,500
14 24,500
15 24,500
16 24,500

45,000
40,000
35,000
30,000 Deman
25,000 Dt
Forecas
20,000 Ft
45,000
40,000
Figure7.5 Moving Average
35,000
30,000 Deman
25,000 Dt
Forecas
20,000 Ft

15,000
10,000
5,000
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
 Figure7.5 Moving Average

Demand
Dt
Forecast
Ft
 Figure7.5 Moving Average

Demand
Dt
Forecast
Ft

6 17
 Figure7.6 Smoothing

Tahoe Salt Forecasts Using Simple Exponential Smoothing

Demand Forecast Absolute Error Mean Squared Error
Period t Dt Level Lt Ft Error Et At MSEt MADt % Error MAPEt TSt
0 22,083
1 8,000 20,675 22,083 14,083 14,083 198,340,278 14,083 176 176 1.00
2 13,000 19,908 20,675 7,675 7,675 128,622,951 10,879 59 118 2.00
3 23,000 20,217 19,908 -3,093 3,093 88,936,486 8,284 13 83 2.25
4 34,000 21,595 20,217 -13,783 13,783 114,196,860 9,659 41 72 0.51
5 10,000 20,436 21,595 11,595 11,595 118,246,641 10,046 116 81 1.64
6 18,000 20,192 20,436 2,436 2,436 99,527,532 8,777 14 70 2.15
7 23,000 20,473 20,192 -2,808 2,808 86,435,714 7,925 12 62 2.03
8 38,000 22,226 20,473 -17,527 17,527 114,031,550 9,125 46 60 -0.16
9 12,000 21,203 22,226 10,226 10,226 112,979,315 9,247 85 62 0.95
10 13,000 20,383 21,203 8,203 8,203 108,410,265 9,143 63 63 1.86
11 32,000 21,544 20,383 -11,617 11,617 110,824,074 9,368 36 60 0.58
12 41,000 23,490 21,544 -19,456 19,456 133,132,065 10,208 47 59 -1.38
13 23,490 11,538 12761
14 23,490
15 23,490
16 23,490

alpha 0.1

45,000
40,000
35,000
30,000
25,000 De
20,000
Fo

15,000
40,000
35,000 Figure7.6 Smoothing

30,000
25,000 De
20,000
Fo

15,000
10,000
5,000
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
 Figure7.6 Smoothing

Demand
Forecast
 Figure7.6 Smoothing

Demand
Forecast

15 16
 Figure7.7 Holt's

Forecasts Using Trend-Corrected Exponential Smoothing (Holt's Model)

Demand Trend Forecast Absolute Error
Period t Dt Level Lt Tt Ft Error Et At Mean Squared Error MSEt MADt % Error MAPEt TSt
0 12,015 1,549
1 8,000 13,008 1,438 13,564 5,564 5,564 30,958,096 5,564 70 70 1
2 13,000 14,301 1,409 14,445 1,445 1,445 16,523,523 3,505 11 40 2
3 23,000 16,439 1,555 15,710 -7,290 7,290 28,732,318 4,767 32 37 0
4 34,000 19,594 1,875 17,993 -16,007 16,007 85,603,146 7,577 47 39.86 -2.15
5 10,000 20,322 1,645 21,469 11,469 11,469 94,788,701 8,355 115 54.83 -0.58
6 18,000 21,570 1,566 21,967 3,967 3,967 81,613,705 7,624 22 49.36 -0.11
7 23,000 23,123 1,563 23,137 137 137 69,957,267 6,554 1 42.39 -0.11
8 38,000 26,018 1,830 24,686 -13,314 13,314 83,369,836 7,399 35 41.48 -1.90
9 12,000 26,262 1,513 27,847 15,847 15,847 102,010,079 8,338 132 51.54 0.22
10 13,000 26,298 1,217 27,775 14,775 14,775 113,639,348 8,981 114 57.75 1.85
11 32,000 27,963 1,307 27,515 -4,485 4,485 105,137,395 8,573 14 53.78 1.41
12 41,000 30,443 1,541 29,270 -11,730 11,730 107,841,864 8,836 29 51.68 0.04
13 31,985
14 33,526
15 35,067
16 36,609

alpha 0.1
Beta 0.2

45,000
40,000
35,000
30,000
25,000
20,000
40,000
35,000
Figure7.7 Holt's

30,000
25,000
20,000
15,000
10,000
5,000
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
 Regression 7.7

HOLT'S MODEL REGRESSION SUMMARY OUTPUT

Regression Statistics
Multiple R 0.4813272
R Square 0.23167587
Adjusted R 0.15484346
Standard E 10666.8834
Observatio 12

ANOVA
df SS MS F Significance F
Regression 1 343092657.343 343092657.3 3.01534 0.113127
Residual 10 1137824009.32 113782400.9
Total 11 1480916666.67

Coefficients Standard Error t Stat P-value Lower 95% Upper 95%Lower 95.0%
Upper 95.0%
Intercept 12,015 6565.01289356 1.830179424 0.097147 -2612.611 26642.91 -2612.611 26642.91
X Variable 1,549 892.009599389 1.73647352 0.113127 -438.5705 3536.473 -438.5705 3536.473

b, estimate of demand
and level at t=0

a, estimate of trend at
t=0
 Figure7.8 Winter

Forecasts Using Trend and Seasonality Corrected Exponential Smoothing)

Demand Seasonal Factor Forecast Mean Squared Absolute Error
Period t Dt Level Lt Trend Tt St Ft Error Et Error MSEt At MADt % Error MAPEt
18,439 524
1 8,000 18,763 484 0.47 8,944 944 892,044 944 944 12 11.81
2 13,000 19,224 479 0.68 13,153 153 457,771 153 549 1 6.49
3 23,000 19,698 478 1.17 23,067 67 306,692 67 388 0 4.43
4 34,000 20,201 483 1.66 33,582 -418 273,730 418 396 1 3.63
5 10,000 20,757 498 0.47 9,663 -337 241,711 337 384 3 3.58
6 18,000 21,766 600 0.68 14,510 -3,490 2,231,081 3,490 902 19 6.21
7 23,000 22,094 546 1.17 26,177 3,177 3,354,300 3,177 1,227 14 7.30
8 38,000 22,657 549 1.67 37,725 -275 2,944,473 275 1,108 1 6.48
9 12,000 23,454 599 0.47 10,841 -1,159 2,766,681 1,159 1,114 10 6.83
10 13,000 23,552 498 0.68 16,420 3,420 3,659,894 3,420 1,344 26 8.78
11 32,000 24,379 564 1.17 28,148 -3,852 4,675,999 3,852 1,572 12 9.07
12 41,000 24,910 558 1.67 41,563 563 4,312,718 563 1,488 1 8.43
13 0.47 12,010
14 0.67 17,427
15 1.18 31,490
16 1.66 45,167

alpha 0.1 L
Beta 0.2 T
Gamma 0.1 S

50,000
45,000 45,167
40,000
35,000
30,000 31,490
Demand
25,000 Forecast
20,000
17,427
15,000
12,010
 Figure7.8 Winter

Bias TSt

944.5 1.00
1,097.8 2.00
1,165.1 3.00
747.0 1.89
409.9 1.07
(3,079.8) (3.42)
97.2 0.08
(177.9) (0.16)
(1,337.3) (1.20)
2,083.0 1.55
(1,768.9) (1.13)
(1,206.2) (0.81)

45,167

31,490
Demand
Forecast

,427
0
 deseasonalized

Forecasts Using Trend and Seasonality Corrected Exponential Smoothing)

Deseasonalizing Demand
Deseasonalized Deseasonalized
Period Demand Demand Demand Seasonal Factor Estimate
t Dt (Eqn 7.2) D t
(Eqn 7.3) D t (Eqn 7.5) S
t
S
(Eqn 7.6) i
1 8,000 18,963 0.42 0.47
2 13,000 19,487 0.67 0.68
3 23,000 19,750 20,010 1.15 1.17
4 34,000 20,625 20,534 1.66 1.66
5 10,000 21,250 21,058 0.47
6 18,000 21,750 21,582 0.83
7 23,000 22,500 22,106 1.04
8 38,000 22,125 22,629 1.68
9 12,000 22,625 23,153 0.52
10 13,000 24,125 23,677 0.55
11 32,000 24,201 1.32
12 41,000 24,725 1.66

45,000

40,000

35,000

30,000

25,000

20,000

15,000

10,000

5,000

0
1 2 3 4 5 6 7 8 9
 Regression

WINTER'S MODEL REGRESSION SUMMARY OUTPUT

Regression Statistics
Multiple R 0.9580652
R Square 0.917889
Adjusted R Squar 0.9042038
Standard Error 414.50331
Observations 8

ANOVA
df SS MS F Significance F
Regression 1 11523809.5 11523809.5 67.07182 0.000179
Residual 6 1030877.98 171812.996
Total 7 12554687.5

CoefficientsStandard Error t Stat P-value Lower 95% Upper 95%Lower 95.0%

Upper 95.0%
Intercept 18,439 440.808708 41.8299089 1.249E-08 17360.37 19517.61 17360.37 19517.61
X Variable 1 524 63.9592497 8.18973841 0.000179 367.3068 680.3123 367.3068 680.3123

initial estimate of
level

initial estimate of
trend
Forecast Errors for Tahoe Salt Forecasting

TS Range
Forecasting Method MAD MAPE(%) Min Max
Four-period moving average 9,719 49 -1.52 2.21
Simple exponential smoothing 10,208 59 -1.38 2.25
Holt's model 8,836 52 -2.15 2.00
Winter's model 1,488 8 #REF! #REF!