Forecast Horizons in Operation Planning

  Supply Chain Management Demand Forecas8ng in a Supply Chain Dr. Eng. Muhammad Rusman, ST.,MT.

  1 Introduc@on to forecas@ng

  • What is forecas@ng?
    • – Primary Func@on is to Predict the

  Future using (@me series related or other) data we have in hand

  • Why are we interested?
    • – Affects the decisions we make today

  • Demand forecas8ng is one of the most fundamental task that a business must perform

  2 Supply Chain Management – @uh 2015

What Makes a Good Forecast?

  • It should be @mely
  • It should be as accurate as possible
  • It should be reliable
  • It should be in meaningful units
  • It should be presented in wri@ng
  • The method should be easy to use and understand in most cases.

  3 Supply Chain Management – @uh 2015 Forecast Horizons in Operation Planning

  4 Supply Chain Management – @uh 2015 Supply Chain Management – @uh 2015

SCM LG EIN Process

  Inventory Company Level Start Coun@ng Company Level for Dom Start Coun@ng Warehouse Level

  Factory (Dom Produc8on)

RDC

CDC DEALER

  Import Start Coun@ng Company Level for

  Import

Supply chain Distribu8on process

  1.  Lead 8me: Loading

  2.  Transporta8on from Factory to CDC Factory Dealers

  St e

  1.  Lead time: Unloading, Loading, p Factory RDC

  Queue CDC Move, 1 x

  2.  Handling cost: Saving: Dealers Operator, Tools/Equip, WH §  Reduce lead @me unloading & loading at CDC (2-5 days) space, Defect risk §  Eliminate transporta@on cost from factory to CDC

  3.  Inv. carrying cost §  Eliminate handling cost at CDC

  4.  Transportation from CDC to Dealers CDC §  Avoid high inventory/aging risk at CDC and RDC

  RDC

  1.  Lead time: Unloading, Loading, Queue

  Move,

  2.  Handling cost: Operator, Tools/Equip, WH space, Defect risk

  RDC St

  3.  Inv. carrying cost e p

  4.  Transportation from RDC to Dealers/

  2 Factory RDC Dealers CDC Wholesaler x x

  Saving: §  Reduce lead @me unloading & loading at CDC (2-5 days) and RDC (2-3 days) §  Eliminate transporta@on cost: from factory to CDC and from CDC to Dealers

  §  Eliminate handling cost at CDC and RDC Dealer/Wholesaler

  §  Avoid high inventory/aging risk at CDC and RDC REQUIREMENT: ü Strong Forecast Accuracy by Weekly S&OP with BM ü Align Prod. Plan to Cust. Demand with “Ship To” code for each branches. ü Sales leveling

  • The methods mostly are theory-based which involved mathema@cal models with parameters that must be calibrated.
  • With liYle or no historical dataà The Delphi method
  • The Delphi Method
    • –  Rely on experts qualita@ve assessment or ques@onnaires o develop forecast
    • –  Individual opinions are compiled and considered. These are anonymously shared among group. Then opinion request is repeated un@l an overall group consensus is (hopefully) reached.

  Supply Chain Management – @uh 2015 Subjective Forecasting Methods

  9 Supply Chain Management – @uh 2015 Demand process

  • Trends, demand consistently increase or decrease over @me
  • Seasonality, Demand shows peaks and valleys at consistent intervals
  • Product life cycle, demand goes through phases of rapid growth, maturity and decline

  10 Supply Chain Management – @uh 2015

  11 Supply Chain Management – @uh 2015

  12 Pola Data

  • Pola horisontal (H) terjadi bilamana data berfluktuasi disekitar nilai rata-rata yg konstan. Suatu produk yg penjualannya tdk meningkat atau menurun selama waktu tertentu.
  • Pola musiman (S) terjadi bilamana suatu deret dipengaruhi oleh faktor musiman (misalnya kuartal tahun tertentu, bulanan, atau hari-hari pada minggu tertentu).

  13 Supply Chain Management – @uh 2015 Pola Data

  • Pola siklis (C) terjadi bilamana datanya dipengaruhi oleh fluktuasi ekonomi jangka panjang seper@ yang berhubungan dengan siklus bisnis.
  • Pola trend (T) terjadi bilamana terdapat kenaikan atau penurunan sekuler jangka panjang dalam data.

  14 Supply Chain Management – @uh 2015 Supply Chain Management – @uh 2015 Moving Average

  • The moving average method calculates the average amount of demand over given @me period and uses this average to predict the future demand

  t

  1

  1 =

  y D t i

  ∑ N i t N = −

  Let D , D , . . . D , . . . be the past values of the

  1 2 n series to be predicted

  Let y = forecast made in period t for the demand in period t t,

  16 Supply Chain Management – @uh 2015 Example

Month Demand Month Demand

  January

  89 July 223 February

  57 August 286 March 144 September 212 April 221 October 275 May 177 November 188 June 280 December 312 3 month MA: (oct+nov+dec)/3=258.33 6 month MA: (jul+aug+…+dec)/6=249.33 12 month MA: (Jan+feb+…+dec)/12=205.33

  17 Supply Chain Management – @uh 2015

Example

  Period (t) Sales (y)

  1

  5.3

  2

  4.4

  3

  5.4

  4

  5

  5.6

  6

  4.8

  (in millions of dollars)

  7

  5.6

  8

  5.6

  9

  5.4

  10

  6.5

  11

  5.1

  12

  5.8

  13

  5

  14

  6.2

  15

  5.6

  16

  6.7

  17

  5.2

  18

  5.5

  19

  5.8

  20

  5.1

  21

  5.8

  22

  6.7

  23

  5.2

  24

  6

  25

  5.8

  18 Supply Chain Management – @uh 2015

  • Use a three-week moving average (k=3) for the department store sales to forecast for the week 24 and 26.
  • The forecast error is 9 .
    • =
    • =

  24

  26 =

  25

  24

  23

  3 ˆ

  5

  6 8 .

  5

  3

2 .

  5

  Supply Chain Management – @uh 2015 Example

  24 = − = − = y y e

  24

  6 ˆ

  5

  y y y y 1 . 9 .

  24

=

  23

  22

  21

  3 ) ( ˆ

  5

  6 2 .

  5 7 .

  3

8 .

  5

  Supply Chain Management – @uh 2015 Example

  • The forecast for the week 26 is 7 .

  y y y y

  • =
  • =
Summary of Moving Averages

  • Advantages of Moving Average Method
    • –  Easily understood
    • –  Easily computed
    • –  Provides stable forecasts

  • Disadvantages of Moving Average Method
    • –  Requires saving lots of past data points: at least the N periods used in the moving average computa@on
    • –  Lags behind a trend
    • –  Ignores complex rela@onships in data

  21 Supply Chain Management – @uh 2015 Exponential Smoothing Method

  • Exponen@al smoothing is the technique that uses a weighted moving average of the past data as the basis for the forecast.

  α α

  • y = D (1 − ) y

  − − t t 1 t

  1

  • α is the weight placed on the demand observa@on and 1- α is the weight placed on last forecast

  where 0 < α < 1 and generally is small for stability of forecasts ( around .1 to .2)

  22 Supply Chain Management – @uh 2015 Exponential Smoothing Method

  In symbols: y = α D + (1 - α ) y

  t+1 t t

  = α D + (1 - α + (1 - α ) ) (α D ) y

  t t-1 t-1

  

2

= α D + (1 - α + (1 - α + . . .

  )(α )D ) (α )D

  t t-1 t - 2

  • Hence the method applies a set of exponen@ally declining weights to past data. It is easy to show that the sum of the weights is exactly one.

   {Or : y = y - D ) } - α (y t + 1 t t t

  23 Supply Chain Management – @uh 2015 Effect of α value on the Forecast

  α

  means that the forecasted

  • Small values of value will be stable (show low variability

  α

  increases the lag of the forecast to the actual

  • – Low data if a trend is present

  α

  mean that the forecast will

  • Large values of more closely track the actual @me series

  24 Supply Chain Management – @uh 2015

Example

  • Jan 23.3
  • Feb 72.3
  • Mar 30.3
  • Apr 15.5
  • And the January Forecast was: 25
  • Using α = .15
    • (1- α)F
      • –  Forecast for Feb: αD jan

    • (1- α)F
      • –  Forecast for Mar: αD feb

    • (1- α)F
      • –  Apr: αD mar

    • (1- α)F
      • –  May: αD apr

    • = + −

  jan

= .15*23.3 + (.85)*25 = 24.745

  11

  8

  47

  43.14

  44.42

  9

  56

  44.30

  45.71

  10

  52

  47.81

  50.85

  55

  43.20

  49.06

  51.42

  12

  54

  50.84

  53.21

  13

  51.79

  53.61

  1 (1 ) t t t y D y

  α α

  Hitung periode 12 dan 13

  45.84

  43

  feb = .15*72.3 + (.85)*24.745 = 31.88

  37.9

  mar = .15*30.3 + .85*31.88 = 31.64

  apr = .15*15.5 + .85*31.64 = 29.22

  25 Supply Chain Management – @uh 2015 Example

  26 Period Demand Forecast , y t+1

  α =0.3 α =0.5

  1 37 - -

  2

  40

  37

  37

  3

  41

  38.5

  Supply Chain Management – @uh 2015

  4

  37

  38.83

  39.75

  5

  45

  38.28

  38.37

  6

  50

  40.29

  41.68

  7

  Valida@on

  • Aker the model specified, its performance characteris@cs should be verified or validated by comparison of its forecast with historical data for the process it was designed to forecast.
  • We can use the error measures such as MAPE

  (Mean absolute percentage error), MSE (Mean square error) or RMSE (Root mean square error)

  27 Supply Chain Management – @uh 2015 n

  1 e i

  .100% MAPE =

  ∑ n 1 y i = i n

  1

  2 MSE e = i

  ∑ n i

  =

  1 RMSE MSE =

  28 Supply Chain Management – @uh 2015

  • Similari@es
    • – Both methods are appropriate for sta@onary series
    • – Both methods depend on a single parameter
    • – Both methods lag behind a trend
    • – One can achieve the same distribu@on of forecast error by seong: α = 2/ ( N + 1) or N = (2 - α)/ α

  Supply Chain Management – @uh 2015 Comparison of MA and ES

  • Differences
    • – ES carries all past history (forever!)
    • – MA eliminates “bad” data aker N periods
    • – MA requires all N past data points to compute new forecast es@mate while ES only requires last forecast and last observa@on of ‘demand’ to con@nue

  Supply Chain Management – @uh 2015 Comparison of MA and ES

  • In linear regression, the model specifica@on assumes that the independent variable, Y, is linear combina@on of the independent variables.
    • = − =

  2

  β =

  S β β

  2 n i i xy xx n D n S

  1 1 ( 1)

  1

  1

  ⎟ ⎜ ⎟ ⎝ ⎠ ⎝ ⎠

  S ⎛ ⎞ ⎛ ⎞ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜

  4 xx n n n n n

  6

  1

  1 ( 1)(2

  2

  2

  ( ) ( )

  • = −

  ∑ ∑ ( )

  = = ⎛ ⎞ ⎜ ⎟ ⎝ ⎠

  2 n n xy i i i i n n S n iD D

  1 ( 1)

  1

  32 ( )

  Supply Chain Management – @uh 2015 Linear Regression

  X β β = +

  31

  Supply Chain Management – @uh 2015 Linear Regression

  ∑

  • = −
  • A hospital receives regular shipments of liquefied oxygen, which it converts to oxygen gas that is used for the life support. The company that sells the oxygen to hospital wishes to forecast the amount of the liquefied oxygen the hospital will use tomorrow. The number of liters of liquefied oxygen used by the hospital in each of the past 30 days is reported in the oxygen.xlsx.

  Supply Chain Management – @uh 2015 Case

  a.  Using a moving average with N=7, forecast tomorrow’s demand b.  Using single exponen@al smoothing with α=0.1, forecast tomorrow’s demand

  33 Supply Chain Management – @uh 2015

  34 Day Demand 1 4804.9 2 4285.0 3 3764.6 4 2486.8 5 3012.2 6 2896.9 7 1985.1 8 3437.0 9 3345.7 10 1841.3

  11 2114.6 12 1803.6 13 2678.7 14 2070.5 15 2645.5 16 3292.6 17 3844.0 18 4901.8 19 3206.5 20 3362.6

  21 2466.2 22 1048.5 23 1431.3 24 2574.3 25 3310.7 26 4415.4 27 2919.6 28 3905.5 29 1332.8 30 1969.5 Supply Chain Management – @uh 2015 End

  35