Table 7.

Table 7.1 Quarterly Demand for Tahoe Salt

Period Demand
Year, Qtr t Dt
01,2 1 8,000
01,3 2 13,000
01,4 3 23,000
02,1 4 34,000
02,2 5 10,000
02,3 6 18,000
02,4 7 23,000
03,1 8 38,000
03,2 9 12,000
03,3 10 13,000
03,4 11 32,000
04,1 12 41,000

Figure 7.1 Quarterly Demand at Tahoe Salt

45,000
40,000
35,000
30,000
Demand

25,000
20,000
15,000
10,000
5,000
0
01,2 01,3 01,4 02,1 02,2 02,3 02,4 03,1 03,2 03,3 03,4 04,1

Year, Quarter
 Figure 7-2 & 7-3

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 Salt

45,000
40,000
35,000
30,000
Demand

25,000
20,000
15,000
10,000
5,000
0
0 2 4 6 8 10 12 14

Period, t
 Regression-1

REGRESSION SUMMARY OUTPUT

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

ANOVA
df SS MS F Significance F
Regression 1 11523809.5238 11523810 67.07182 0.0001786086
Residual 6 1030877.97619 171813
Total 7 12554687.5

Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 18,439 440.808707878 41.82991 1.25E-08 17360.367255 19517.609 17360.36726 19517.60894
X Variable 1 524 63.9592496814 8.189738 0.000179 367.30676332 680.31228 367.3067633 680.3122843

Initial Level, L
Trend, T
 Figure7-4

Deseasonalized
Period Demand Demand Seasonal Factor Estimate
t Dt (Eqn 7.4) Dt (Eqn 7.5) (Eqn 7.6) SForecast
i
1 8,000 18,963 0.42 St 0.47 8,913
2 13,000 19,487 0.67 0.68 13,251
3 23,000 20,011 1.15 1.17 23,413
4 34,000 20,535 1.66 1.67 34,293
5 10,000 21,059 0.47 9,898
6 18,000 21,583 0.83 14,676
7 23,000 22,107 1.04 25,865
8 38,000 22,631 1.68 37,794
9 12,000 23,155 0.52 10,883
10 13,000 23,679 0.55 16,102
11 32,000 24,203 1.32 28,318
12 41,000 24,727 1.66 41,294

Forecasted Data
45,000
Forecasted
Period Demand 40,000
Year, Qtr t Ft 1 35,000
03,2 13 11,868 30,000 Demand
03,3 14 17,527 Dt
03,4 15 30,770 25,000
04,1 16 44,794 20,000 Deseasonali
Demand
15,000 (Eqn 7.4)
10,000
5,000
0
1 2 3 4 5 6 7 8 9 10 11 12
 Demand Squared Absolute
Period t Dt Level Lt Forecast Ft Error Et Error Errror At % Error
0 2,017.9
1 2,024.0 2,019.8 2,017.9 -6.1 37.2 6.1 0.3%
2 2,076.0 2,037.8 2,019.8 -56.2 3,152.9 56.2 2.7%
3 1,992.0 2,023.2 2,037.8 45.8 2,097.2 45.8 2.3%
4 2,075.0 2,039.7 2,023.2 -51.8 2,687.5 51.8 2.5%
5 2,070.0 2,049.4 2,039.7 -30.3 916.4 30.3 1.5%
6 2,046.0 2,048.3 2,049.4 3.4 11.6 3.4 0.2%
7 2,027.0 2,041.5 2,048.3 21.3 454.3 21.3 1.1%
8 1,972.0 2,019.3 2,041.5 69.5 4,830.6 69.5 3.5%
9 1,912.0 1,985.0 2,019.3 107.3 11,511.1 107.3 5.6%
10 1,985.0 1,985.0 1,985.0 0.0 0.0 0.0 0.0%
2,017.9 103 2,570 39.2 2.0%
α= 0.32
Select smoothing constant by Minimizing MSE (Figure 7-5)
Use Data | Analysis | Solver with Target Cell set to be F13.
Select smoothing constant by Minimizing MAD (Figure 7-6)
Use Data | Analysis | Solver with Target Cell set to be G13.

Optimal smoothing constant is given in Cell B14

 Figure 7-7

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 Demand
25,000 Dt
20,000 Forecast
15,000 Ft
10,000
5,000
0
0 2 4 6 8 10 12 14

Tahoe Salt Forecasts Using Four-Period Moving Average

 Figure 7-8

Demand Absolute Error Mean Squared Error

Period t Dt Level Lt Forecast 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
23,490 11,538 12,761
23,490
23,490
23,490

α 0.1

45,000
40,000
35,000
30,000
25,000 Demand
20,000 Forecast
15,000
10,000
5,000
0
1 2 3 4 5 6 7 8 9 10 11 12

Tahoe Salt Forecasts Using Simple Exponential Smoothing

 Figure 7-9

Trend Absolute Error Mean Squared Error

Period t Demand Dt Level Lt Tt Forecast Ft Error Et At 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 69.6 69.55 1.00
2 13,000 14,301 1,409 14,445 1,445 1,445 16,523,523 3,505 11.1 40.33 2.00
3 23,000 16,439 1,555 15,710 -7,290 7,290 28,732,318 4,767 31.7 37.46 -0.06
4 34,000 19,594 1,875 17,993 -16,007 16,007 85,603,146 7,577 47.1 39.86 -2.15
5 10,000 20,322 1,645 21,469 11,469 11,469 94,788,701 8,355 114.7 54.83 -0.58
6 18,000 21,570 1,566 21,967 3,967 3,967 81,613,705 7,624 22.0 49.36 -0.11
7 23,000 23,123 1,563 23,137 137 137 69,957,267 6,554 0.6 42.39 -0.11
8 38,000 26,018 1,830 24,686 -13,314 13,314 83,369,836 7,399 35.0 41.48 -1.90
9 12,000 26,262 1,513 27,847 15,847 15,847 102,010,079 8,338 132.1 51.54 0.22
10 13,000 26,298 1,217 27,775 14,775 14,775 113,639,348 8,981 113.7 57.75 1.85
11 32,000 27,963 1,307 27,515 -4,485 4,485 105,137,395 8,573 14.0 53.78 1.41
12 41,000 30,443 1,541 29,270 -11,730 11,730 107,841,864 8,836 28.6 51.68 0.04
31,985
33,526
35,067
36,609

α 0.1
β 0.2

45,000
40,000
35,000
30,000
25,000 Demand
20,000 Forecast
15,000
10,000
5,000
0
1 2 3 4 5 6 7 8 9 10 11 12
 holts-regression

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 343092657.34 3.0153403 0.113127
Residual 10 1137824009.32 113782400.93
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 1.8301794239 0.0971473 -2612.611 26642.914 -2612.611 26642.914
X Variable 1,549 892.009599389 1.73647352 0.113127 -438.5705 3536.4726 -438.5705 3536.4726

L0 , estimate of demand
and level at t=0

T0, estimate of trend at

t=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(Eqn
t 7.3) Dt (Eqn 7.5) (Eqn 7.6) Si
1 8,000 18,963 0.42 St 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 Demand
25,000
20,000 Deseasonl-
15,000 ized Demand
10,000
5,000
0
1 2 3 4 5 6 7 8 9 10 11 12
 winters-regression

WINTER'S MODEL REGRESSION SUMMARY OUTPUT

Regression Statistics
Multiple R 0.95806524
R Square 0.917889
Adjusted R Squar 0.90420383
Standard Error 414.503312
Observations 8

ANOVA
df SS MS F Significance F
Regression 1 11523809.52 11523809.52 67.071815 0.0001786
Residual 6 1030877.976 171812.996
Total 7 12554687.5

CoefficientsStandard Error t Stat P-value Lower 95% Upper 95%Lower 95.0%Upper 95.0%
Intercept 18,439 440.8087079 41.82990891 1.249E-08 17360.367 19517.609 17360.367 19517.609
X Variable 1 524 63.95924968 8.18973841 0.0001786 367.30676 680.31228 367.30676 680.31228

L0 , initial estimate
of level

T0, initial estimate of

trend
 Figure 7-10

Demand Absolute Error Mean Squared

Period t Dt Level Lt Trend Tt Seasonal Factor St Forecast Ft Error Et At Error MSEt MADt % Error MAPEt TSt
18,439 524
1 8,000 18,866 514 0.47 8,913 913 913 832,857 913 11 11.41 1.00
2 13,000 19,367 513 0.68 13,179 179 179 432,367 546 1 6.39 2.00
3 23,000 19,869 512 1.17 23,260 260 260 310,720 450 1 4.64 3.00
4 34,000 20,380 512 1.67 34,036 36 36 233,364 347 0 3.50 4.00
5 10,000 20,921 515 0.47 9,723 -277 277 202,036 333 3 3.36 3.34
6 18,000 21,689 540 0.68 14,558 -3,442 3,442 2,143,255 851 19 5.98 -2.74
7 23,000 22,102 527 1.17 25,981 2,981 2,981 3,106,508 1,155 13 6.98 0.56
8 38,000 22,636 528 1.67 37,787 -213 213 2,723,856 1,037 1 6.18 0.42
9 12,000 23,291 541 0.47 10,810 -1,190 1,190 2,578,653 1,054 10 6.59 -0.72
10 13,000 23,577 515 0.69 16,544 3,544 3,544 3,576,894 1,303 27 8.66 2.14
11 32,000 24,271 533 1.16 27,849 -4,151 4,151 4,818,258 1,562 13 9.05 -0.87
12 41,000 24,791 532 1.67 41,442 442 442 4,432,987 1,469 1 8.39 -0.63
13 0.47 11,940
14 0.68 17,579
15 1.17 30,930
16 1.67 44,928

α 0.05
β 0.1 50,000
γ 0.1
40,000

30,000 Demand
20,000 Forecast

10,000

0
1 2 3 4 5 6 7 8 9 10 11 12
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 1,469 8 -2.74 4.00