SOLUTIONS TO SELECTED PROBLEMS FROM NAHMIAS BOOK

CHAPTER 2
FORECASTING

2.13

Fcst1Fcst2DemandErr1Err2Er1^2Er2^2|Err1|
22321025633461089211633
2893203405120260140051
4303903755515302522555
134112110242576424
19015022535751225562535
5504905252535625122525
1523.51599.16637.16666
(MSE1(MSE2)(MAD1)

Err2 e1/D *100 e2/D 100

4612.8906217.96875
2015.0000
5.88253
1514.666674.00000
221.818181.81818
7515.5555633.33333
354.7619056.66667
32.1666614.1154911.61155
(MAD2)(MAPE1)(MAPE2)

2.14

It means that E(ei) 0. This will show up by considering

n e
i
i 1

A bias is indicated when this sum deviates too far from zero.

2.16

MA(3)forecast:258.33
MA(6)forecast:249.33
MA(12)forecast:205.33

2.17, 2.18, and 2.19.

OnestepaheadTwostepahead
e
e
MonthForecastForecastDemand
1
2

July205.50149.7522317.5073.25
August225.25205.5028660.7580.50
September241.50225.2521229.5013.25
October250.25241.5027524.7533.50
November249.00250.2518861.0062.25
December240.25249.0031271.7563.00
MAD=44.254.3

The one step ahead forecasts gave better results (and should have according to the theory).

2.20

Month Demand

MA(3) MA(6)

July223226.00161.33
August286226.67183.67
September212263.00221.83
October275240.33233.17
November188257.67242.17
December312225.00244.00

MA (6) Forecasts exhibit less variation from period to period.

2.21

An MA(1) forecast means that the forecast for next period is simply the current period's
demand.
Month

Demand

MA(4)

MA(1)

Error

MonthDemandMA(4)MA(1)

Error
July223205.5028057
August286225.2522363
September212241.5028674
October275250.2521263
November188249.0027587
December312240.25188124
MAD=78.0

(Much worse than MA(4))

2.35

a)

V1 = (16 + 32 + 71 + 62)/4 = 45.25

V2 = (14 + 45 + 84 + 47)/4 = 47.5
1. G0 = (V2 - V1)/N = 0.5625
2. S0 = V2 + G0 (N-1/2) = 47.5 + (0.5625)(3/2) = 48.34
3. ct =

Dt
Vi N 1/ 2 j G0

-2N+1 = t 0

c-7 =

16
= 0.36
45.25 5/ 2 1..56

c-6 =

32
= 0.71
45.25 5/ 2 2.56

c-5 =

71
= 1.56
43.25 5/ 2 3.56

c-4 =

62
= 1.35
45.25 5/ 2 4.56

c-3 =

14
= 0.30
47.5 5/ 2 1.56

c-2 =

45
= 0.95
47.5 5/ 2 2.56

c-1 =

84
= 1.76
47.5 5/ 2 3.56

c0 =

47
= 0.97
47.5 5/ 2 4.56

(c7 + c3)/2 = .33

(c6 + c2)/2 = .83
(c5 + c1)/2 = 1.66
(c4 + c0)/2 = 1.16
Sum =

3.98

Norming factor = 4/3.9 = 1.01

Hence the initial seasonal factors are:

b)

c-3 = .33

c-1 = 1.67

c-2 = .83

c-0 = 1.17

= 0.2, = 0.15, = 0.1, D1 = 18

S1 = (D1/c-3) + (1-)(S0 + G0) = 0.2(18/0.33)
+ 0.8(48.34 + 0.56) = 50.03
G1 = (S1 - S0) + (1 - ) = G0 = 0.1(50.03 - 48.34)
+ 0.9(0.56) = 0.70
c1 = (D1/S1) + (1-)c3 = 0.15(18/50.03) + 0.85(0.33)
= .3345

c)

Forecasts for 2nd, 3rd and 4th quarters of 1993

F1,2 = [S1 + G1]c2 = (50 + .70)0.83 = 42.08
F1,3 = [S1 + 2G1]c3 = (50 + 2(.70))1.67 = 85.84
F1,4 = [S1 + 3G1]c4 = (50 + 3(.70))1.17 = 60.96

2.36
Period
1
2
3
4

51
86
66

Forecast
Forecast
from
from
30(d) e
31(c) e
t
t
35.8
82.4
56.5

15.2
3.6
9.5

42.08
85.84
60.96

8.92
0.16
5.04

MAD = 9.43 MAD = 4.71

MSE = 111.42 MSE = 35.00
Hence we conclude that Winter's method is more accurate.

2.37

S1 = 50.03
G1 = 0.67

= 0.2

= 0.15

= 0.1

D1 = 18
D2 = 51
D3 = 85
D4 = 66

S2 = 0.2(51/0.83) + 0.8(50.03 + 0.70) = 52.87

G2 = 0.1(52.87 - 50.03) + 0.9(0.70) = 0.914
S3 = 0.2(86/1.67) + 0.8(52.87 + 0.914) = 53.33
G3 = 0.1(53.33 - 52.85) + 0.9(0.885) = 0.8445
S4 = 0.2(66/1.17) + 0.8(53.33 + 0.8445) = 54.62
G4 = 0.1(54.62 - 53.33) + 0.9(0.8445) = 0.8891
c1 = (.15)[18/50] + (0.85)(.33) = .3345 .34
c2 = (.15)[51/52.85] + 0.85(0.83) = .8502 .85
c3 = (.15)(86/53.29) + 0.85(1.67) = 1.6616 1.66
c4 = (.15)(66/54.59) + 0.85(1.17) = 1.1758 1.18
The sum of the factors is 4.02. Norming each of the factors by multiplying by
4/4.02 = .995 gives the final factors as:
c1 = .34
c2 = .84
c3 = 1.65
c4 = 1.17
The forecasts for all of 1995 made at the end of 1993 are:
F4,9 = [S4 + 5G4]c1 = [54.62 + 5(0.89)]0.34 = 20.08
F4,10 = [S4 + 6G4]c2 = [54.62 + 6(0.89)]0.84 = 50.37
F4,11 = [S4 + 7G4]c3 = [54.62 + 7(0.89)]1.65 = 100.40
F4,12 = [S4 + 8G4]c4 = [54.62 + 8(0.89)]1.17 = 72.24