Computer Science Centre University of Indonesia Inform ation M anagem ent R ESEA R C H G R O U P Forecasting Chapter 15 Management Science, 7th edition Bernard W Taylor III (2002)
Dec 14, 2015
Computer Science CentreUniversity of Indonesia
In fo rm ationM anagem entR E S E A R C H G R O U P
Forecasting
Chapter 15Management Science, 7th edition
Bernard W Taylor III (2002)
In fo rm ationM anagem entR E S E A R C H G R O U P
Agenda
• Intro to Forecasting• Forecasting method• Time Series• Regression & Multiple Regression• Other statistical forecasting method• Tugas untuk 31 Oktober 2003
(presentasi)
In fo rm ationM anagem entR E S E A R C H G R O U P
Intro
• Forecasting is a prediction of what will occur in the future
• Although impossible to predict future exactly, forecast can provide reliable guidelines for decision making
In fo rm ationM anagem entR E S E A R C H G R O U P
Forecast Movement Forms(a)Trend (b) Cycle
(economic)(c) Seasonal (d)Trend & Seasonal
In fo rm ationM anagem entR E S E A R C H G R O U P
Forecasting Methods
• Time series• Regression• Qualitative methods (must read
yourself!)
In fo rm ationM anagem entR E S E A R C H G R O U P
Time Series
• Statistical techniques that make use of historical data
• Assumption: what happen in the past will happen in the future
In fo rm ationM anagem entR E S E A R C H G R O U P
Moving Average
• Tends to smooth the random increase and decrease
• Computed for specific period
915
507513011090
5
5
11
5
i
DMA
In fo rm ationM anagem entR E S E A R C H G R O U P
Weighted Moving Average
• To adjust MA method to reflect more closely recent fluctuation
• Baca sendiri
In fo rm ationM anagem entR E S E A R C H G R O U P
Exponential Smoothing
• Weights most recent data more strongly than distant past data.• Usefull if changes in data are result of an actual change (such as seoasons) rather than just random
change• Rumus:
F = forecastD = actual demand = smoothing constant
• What happens if =0 or =1…?
ttt FDF )1(1
In fo rm ationM anagem entR E S E A R C H G R O U P
Case
F2 = D1 + (1- )F1
= 0,3.37 + 0,7.37 = 37
F3 = D2 + (1- )F2
= + 0,7.37 = 37
In fo rm ationM anagem entR E S E A R C H G R O U P
Adjusted Exponential Smoothing• Exponential smoothing generally lies below the actual demand (especially in upward trends)• Adjusted exponential smoothing adds a certain value to adjust the forecast so it reflects the
actual demand more precisely• Rumus:
T = trend factor = smoothing constant for trend
111 ttt TFAF
tttt TFFT )1()( 11
In fo rm ationM anagem entR E S E A R C H G R O U P
Linear Trend Line
• Use least square regression• Baca sendiri…!
In fo rm ationM anagem entR E S E A R C H G R O U P
Seasonal Adjustment
• We need to adjust seasonality by multiplying the normal forecast by a seasonal factor
In fo rm ationM anagem entR E S E A R C H G R O U P
Example: Turkey Demand
Use linear trend to get forecast for year 5 = 58.17
In fo rm ationM anagem entR E S E A R C H G R O U P
Multiple Regression
• Relationship between a dependent variable and two or more independent variable
• Formulay = ax1+bx2 + … + c
In fo rm ationM anagem entR E S E A R C H G R O U P
Example of Multiple Regression
• Dependant variable: attendance
• Independent variable: wins & promotion
• We can predict attendance if we have $60.000 for promotion and an expeted wins of seven games
Computer Science CentreUniversity of Indonesia
In fo rm ationM anagem entR E S E A R C H G R O U P
Tugas