MENG 445 Production and Inventory Control

Forecasting

Dr. Lotfi K. Gaafar

Production and Inventory Control (1)

Gaafar 2005

Production Planning and Control

The Production Control System
Demand forecasting
Sales and order entry Customer

Aggregate planning

Materials requirement planning

Shop-floor scheduling and control

Production

Shipping and receiving

Inventory management

Inventory

Vendors

Production and Inventory Control (2)

Gaafar 2005

Forecasting
Basic Problem: predict demand for planning purposes. Forecasting Tools: Qualitative: Delphi Analogies Quantitative: Causal models (e.g., regression models) Time series models

Production and Inventory Control (3)

Gaafar 2005

Forecasting Laws
1) Forecasts are always wrong! 2) Forecasts always change! 3) The further into the future, the less reliable the forecast!
40% 20%
+10% -10% Start of season

Trumpet of Doom

16 weeks 26 weeks

Production and Inventory Control (4)

Gaafar 2005

Why Forecasting?
Most decisions made in PP&C are plans for future How much of a certain product should I make next week? How much inventory should I keep in the May 2001? If you forecast the future (more accurately), you will (more likely) have better plans. Many decisions (plans) in PPC depend on the forecasts of demand. Forecasting demand versus forecasting sales.

Production and Inventory Control (5)

Gaafar 2005

Qualitative Forecasting
Delphi Technique: Iteratively tapping multiple experts opinion with continuous comparisons and revisions.

Production and Inventory Control (6)

Gaafar 2005

Quantitative Forecasting
Goals: Predict future from past Smooth out noise Standardize forecasting procedure Methodologies: Causal Forecasting: regression analysis Time Series Forecasting: moving average exponential smoothing regression analysis seasonal models
Gaafar 2005

Production and Inventory Control (7)

The Forecasting Model

Historical Data

Forecast Generation
Current Observation

Forecast

Forecast Control

Managerial Judgment and Experience

Modified Forecast
Production and Inventory Control (8) Gaafar 2005

Time Series Forecasting

Historical Data Forecast

xt, t = 1, ,T
Observed value at time t

Time series model

(T )

xt

Predicted value for time T+ made at time T.

T
Production and Inventory Control (9)

T+

t
Gaafar 2005

Possible Patterns

Source: Stevenson, Stevenson, Nahmias, J. Production and Operations Analysis, 2009.

Production and Inventory Control (10)

Gaafar 2005

The Constant Model

xt

Model: xt = a + et, et NID(0, s2)

Assumptions: No trend, only random noise t Noise is normally distributed Noise is independent Noise has a constant variance (doesnt change with time) Noise averages out to zero
Production and Inventory Control (11) Gaafar 2005

The Constant Model

Moving Average (same as least squares)
1 , x t N t T N 1
T

MT

V (M T ) s , N
2

T (T ) M T 2 1 x

1 s N

Averages the latest N observations. N ranges from 2 to 12. Larger N provides stability, but less responsive. Assigns equal weights to the latest N observations Does not adapt to error
Production and Inventory Control (12) Gaafar 2005

The Constant Model

Moving Average- Example
t 1 2 3 4 5 6 7 8 9 10 xt 20 22 23 20 18 22 22 23 20 24
21.67 21.67 20.33 20.00 20.67 22.33 21.67 22.33 21.67 21.67 20.33 20.00 20.67 22.33 21.67 -1.67 -3.67 1.67 2.00 2.33 -2.33 2.33 1.67 3.67 1.67 2.00 2.33 2.33 2.33

Mt

t (t 1) x

et

|et|

2.29
Production and Inventory Control (13) Gaafar 2005

The Constant Model

Exponential Smoothing
a xT (1 a ) S T 1 , V (S T ) 2 a s

T (T ) S T 2 1 x

a
2 a

Combines current observation and previous forecasts for future predictions. a ranges from 0.1 to 0.5. Smaller a provides stability, but less responsive.

Assigns decaying weights to all observations

Adapts to error
Production and Inventory Control (14) Gaafar 2005

The Constant Model

Exponential Smoothing- Example
t 1 2 3 4 5 6 7 8 9 10 xt 20 22 23 20 18 22 22 23 20 24 St
21.67 21.83 22.42 21.21 19.60 20.80 21.40 22.20 21.10 22.55 21.67 21.83 22.42 21.21 19.60 20.80 21.40 22.20 21.10 1.17 -2.42 -3.21 2.40 1.20 1.60 -2.20 2.90 1.17 2.42 3.21 2.40 1.20 1.60 2.20 2.90 2.14

t (t 1) x

et

|et|

Production and Inventory Control (15)

Gaafar 2005

The Linear Model

xt

Model: xt = a + bt + et, et NID(0, s2) Assumptions: Linear trend Noise is normally distributed t Noise is independent Noise has a constant variance (doesnt change with time) Noise averages out to zero

Production and Inventory Control (16)

Gaafar 2005

The Linear Model

Least Squares

t T N 1 T

t x

t T N 1

2 t (

x t T N 1 t T N 1 T 2 t T N 1

x t t )

t x b a

2 1 ( T t ) (T ) 2 1 ( T ) a b s xT N S tt

Production and Inventory Control (17)

Gaafar 2005

The Linear Model

Double Moving Average

M
M
[ 2] T

1 N
T

t T N 1

x,
t
[ 2]

N 1 E[ M T ] E[ xT ] b 2
( N 1) b 2

1 N

t T N 1

M t,
[ 2]

E[ M T ] E[ xT ] ( N 1)b E[ M T ]

T (T ) 2 M T M T , x

T (T ) (2 x

2 2 [ 2] ) M T (1 ) M T , (point estimate) N 1 N 1

2 2 1 (T t )2 [ 2] T (T ) (2 N 1) M T (1 N 1) M T 2 1 N S tt s x

Production and Inventory Control (18)

Gaafar 2005

The Linear Model

Double Exponential Smoothing
[ 2] T

S
S
T

a S T (1 a ) S T 1 , E[S T
[ 2]
[ 2] [ 2]

[ 2]

2 ] E[ xT ] b E[S T ] b a a
2

[ 2]

a S T (1 a ) S T 1 ,
[ 2]

E[ S T ] E[ xT ]

b E[ S T ]

b a

T (T ) 2 S T S T , x

T (T ) (2 x

a a [ 2] ) S T (1 ) S T , (point estimate )

[ 2] (0) - b(0), S a (0) - 2 b(0) S0 a 0 a a

a a [ 2] 1 (T t )2 T (T ) (2 ) S T (1 ) S T 2 1 N S tt s x
Production and Inventory Control (19) Gaafar 2005

The Seasonal Model

Model:
xt = (a + bt)cL + et, et NID(0, s2)

Example The quarterly sales of company XYZ over the past five years are given below. Use this data to forecast the sales in every quarter of years 6 and 7. Use the same data to forecast the total sales of years 6 and 7.

Quarter Year 1 2 3 4 5 I 17.51 21.33 25.26 27.38 32.99 II 40.24 47.81 50.04 65.50 67.27 III 35.76 45.45 53.92 56.58 65.94 IV 20.08 24.11 29.65 26.51 34.34
Gaafar 2005

Production and Inventory Control (20)

The Seasonal Model

Example
70.00 60.00 50.00 40.00 30.00 20.00 10.00 0.00
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

Production and Inventory Control (21)

Gaafar 2005

The Seasonal Model

Example
Year 1 2 3 4 5 I 17.51 21.33 25.26 27.38 32.99 II 40.24 47.81 50.04 65.50 67.27
Year 1 2 I 0.62 0.62 0.64 0.62 0.66 0.63

III 35.76 45.45 53.92 56.58 65.94

II 1.42 1.38 1.26 1.49 1.34 1.38

IV 20.08 24.11 29.65 26.51 34.34

III 1.26 1.31 1.36 1.29 1.32 1.31

28.40 34.67 39.72 43.99 50.14

IV 0.71 0.70 0.75 0.60 0.68 0.69

a = 23.55 b = 5.28.

Seasonal Indices

3 4 5 Average

Production and Inventory Control (22)

Gaafar 2005

The Seasonal Model

Forecasted Values
Year 6 average = 23.55 + 6 * 5.28 = 55.23 Year 7 average = 23.55 + 7 * 5.28 = 60.51
Quarter Index I 0.63 II 1.38 III 1.31 IV 0.69

Seasonal Indices

Year

I 34.77 38.09

II 76.05 83.33

III 72.11 79.00

IV 37.95 41.58

Forecasted Values

6 7

Production and Inventory Control (23)

Gaafar 2005