introduction to operations management forecasting (ch.3) hansoo kim ( 金翰秀 ) dept. of...
TRANSCRIPT
Introduction to Operations Management
Forecasting (Ch.3)Forecasting (Ch.3)
Hansoo Kim ( 金翰秀 )Dept. of Management Information
Systems, YUST
OM Overview
Class Overview(Ch. 0)
Project Management
(Ch. 17)
Strategic Capacity Planning(Ch. 5, 5S)
Operations, Productivity, and Strategy
(Ch. 1, 2)
Mgmt of Quality/Six Sigma Quality
(Ch. 9, 10)
Supply Chain Management
(Ch 11)
Location Planning and Analysis
(Ch. 8)
Demand MgmtForecasting
(Ch 3)
Inventory Management
(Ch. 12)
Aggregated Planning
(Ch. 13)
Queueing/ Simulation
(Ch. 18)
MRP & ERP (Ch 14)
JIT & Lean Mfg System
(Ch. 15)
Term Project
Process Selection/
Facility Layout; LP(Ch. 6, 6S)X X X X X
XX X X
Today’s Outline What is Forecasting? 수요예측이란 ? Types of Forecasting 종류 Seven Steps of Forecasting 절차 Qualitative vs. Quantitative Forecasts
정성적 /정량적 방법 Various Forecasting Methods
여러 가지 수요예측 방법들 Evaluation Measures 평가방법
Terms( 용어 ) Associative models (联合模型) Bias (偏差) Centered moving average (中心
滑动平均数) Control chart (控制图) Correlation (想关系数) Cycles (循环变动) Delphi method (德尔菲法) Error (误差) Exponential smoothing (指数平滑
法) Forecast ( 预册 ) Irregular variation (不规则变化) Judgmental forecasts (通过判断作
出预测) Least square line (最小二乘直线)
Linear trend equation (线形趋势模型)
Mean absolute deviation, MAD (绝对平均偏差)
Mean squared error, MSE (标准偏差)
Moving average (滑动平均法) Naive forecast (简单预测法) Predictor variable (预测变动) Random variations (随机变量) Regression (回归分析) Seasonal variations (季节性变
动) Time series (时间序列) Tracking signal (跟踪信号) Trend (长期趋势变动) Trend-adjusted exponential
smoothing (调整长期趋势后的指数平滑法)
Example: ForecastingPast Demand for Chairs, Smart Furniture
0
500
1000
1500
2000
2500
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Week
Dem
and
Demand > Products --- Lost SalesDemand < Products --- Inventory cost
What is demand in 21st week?
Week Demand1 8002 14003 10004 15005 15006 15007 13008 18009 1700
10 130011 170012 170013 150014 230015 230016 200017 170018 180019 220020 190021 ???
What is Forecasting?
The art and science of predicting future events, Using historical data
Underlying basis of all business decisions Production Inventory Personnel Facilities
Sales will be $200 Million!
CRM: Customer Relation Management
Accounting Cost/profit estimates
Finance Cash flow and funding
Human Resources Hiring/recruiting/training
Marketing Pricing, promotion, strategy
MIS IT/IS systems, services
Operations Schedules, MRP, workloads
Product/service design New products and services
Uses of Forecasts
Types of Forecasts by Time Horizon
Short-range forecast ( 단기 예측 ) Up to 1 year; usually less than 3 months Job scheduling, worker assignments
Medium-range forecast ( 중기 예측 ) 3 months to 3 years Sales & production planning, budgeting
Long-range forecast ( 장기 예측 ) 3+ years New product planning, facility location, R&D
Influence of Product Life Cycle
Stages of introduction and growth require longer forecasts than maturity and decline
Forecasts useful in projecting staffing levels, inventory levels, and factory capacity
as product passes through life cycle stages
Introduction, Growth, Maturity, Decline( 도입기 , 성장기 , 성숙기 , 쇠퇴기 )
Strategy and Issues During a Product’s Life
Steps in the Forecasting Process
Step 1 Determine purpose of forecast
Step 2 Establish a time horizon
Step 3 Select a forecasting technique
Step 4 Obtain, clean and analyze data
Step 5 Make the forecast
Step 6 Monitor the forecast
“The forecast”
Realities of Forecasting
Forecasts are seldom perfect Forecasts are seldom perfect (( 수요예측은 잘 맞지 않는다수요예측은 잘 맞지 않는다 !!)!!)
Most forecasting methods assume that there is some underlying stabilitystability in the system ( 신제품 보다는 성숙기에 도달한 제품 )
Both product family and aggregated product family and aggregated product product forecasts are more accurate than individual product forecasts ( 단일제품 보다는 제품群으로 예측하는 것이 더 정확하다 )
Forecasting Approaches
Used when situation is ‘stable’ & historical data exist Existing products Current technology
Involves mathematical techniques e.g., forecasting sales of
color televisions
Quantitative Methods( 정량적방법 )
Used when situation is vague & little data exist New products New technology
Involves intuition, experience e.g., forecasting sales on
Internet
Qualitative Methods( 정성적방법 )
Overview of Qualitative Methods ( 정성적인 방법들 ) Jury of executive opinion
Pool opinions of high-level executives, sometimes augment by statistical models
Delphi method ( 델파이 기법 )** Panel of experts, queried iteratively
Sales force composite Estimates from individual salespersons are
reviewed for reasonableness, then aggregated
Consumer Market Survey Ask the customer
Overview of Quantitative Approaches
Naïve approach Moving averages Exponential
smoothing Trend projection
Linear regression
Time-series Models
Associative models
What is a Time Series?
Set of evenly spaced numerical data Obtained by observing response variable at
regular time periods Forecast based only on past values
Assumes that factors influencing past and present will continue influence in future
Example Year: 2000 2001 2002 2003 2004 Sales: 78.7 63.5 89.7 93.2 92.1
Product Demand Charted over 4 Years with Trend and Seasonality
Year1
Year2
Year3
Year4
Seasonal peaks Trend component
Actual demand line
Average demand over four years
Dem
and
for p
rodu
ct o
r ser
vice
Random variation
Time Series Components
TrendTrend
SeasonalSeasonal
CyclicalCyclical
RandomRandom
Overview of Qualitative Methods ( 정량적인 방법들 )
Naïve Approach Moving Average (MA) 滑动平均法 Weighted Moving Average Exponential Smoothing Method
(指数平滑法 ) Exponential Smoothing with Trend
Adjustment (调整长期趋势后的指数平滑法 )
Naive Approach (简单预测法 )
Assumes demand in next period is the same as demand in most recent period e.g., If May sales were
48, then June sales will be 48
Sometimes cost effective & efficient
© 1995 Corel Corp.
Moving Average (MA) Method滑动平均法
MA is a series of arithmetic means Used if little or no trend Used often for smoothing
Provides overall impression of data over time
Equation
MAMAnn
nn Demand inDemand in PreviousPrevious PeriodsPeriods
Moving Average
whereFt = Forecast for time period tMAn = n period moving averageAt – 1 = Actual value in period t – 1n = Number of periods (data points) in the moving averageFor example, MA3 would refer to a three-period moving average forecast, and MA5 would refer to a five-period moving average forecast.
Moving Average Example
You’re manager of a museum store that sells historical replicas. You want to forecast sales for 2008 using a 3-period moving average.
2003 42004 62005 52006 32007 7
© 1995 Corel Corp.
Moving Average Solution
Time Response Yi
Moving Total (n=3)
Moving Average
(n=3) 2003 4 NA NA 2004 6 NA NA 2005 5 NA NA 2006 3 4+6+5=15 15/3=5.0 2007 7 6+5+3=14 14/3=4.7 2008 NA 5+3+7=15 15/3=5.0
Weighted Moving Average Method
Used when trend is present Older data usually less important
Weights based on intuitionOften lay between 0 & 1, & sum to 1.0
Equation
WMA =WMA =ΣΣ(Weight for period (Weight for period nn) (Demand in period ) (Demand in period nn))
ΣΣWeightsWeights
Actual Demand, Moving Average, Weighted Moving Average
0
5
10
15
20
25
30
35
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month
Sal
es D
eman
d
Actual sales
Moving average
Weighted moving average
Disadvantages of Moving Average Methods
Increasing n makes forecast less sensitive to changes
Do not forecast trend well Require much historical
data
Measure for Forecasting Error
Mean Square Error (MSE,标准偏差 )
Mean Absolute Deviation (MAD,绝对平均偏差 )
Mean Absolute Percent Error (MAPE)
2
n
1i
2ii
n
errorsforecast
n
)y(yMSE
nn
yyMAD
n
iii
|errorsforecast |
|ˆ|1
n
actual
forecastactual
100MAPE
n
1i i
ii
Exponential Smoothing Method (指数平滑法 )
Form of weighted moving average Weights decline exponentially Most recent data weighted most
Requires smoothing constant () Ranges from 0 to 1 Subjectively chosen
Involves little record keeping of past data
Exponential Smoothing Equations
Ft = At -1 + (1- )At -2 + (1- )2·At - 3 + (1- )3At - 4 + ... + (1- )t-1·A0
Ft = Forecast value
At = Actual value
= Smoothing constant
FFtt = = FFtt-1-1 + + ((AAtt-1-1 - - FFtt-1-1) = ) = AAtt-1-1+(1- +(1- ) ) FFtt-1-1
Use for computing forecast
1
1
0
)1(
it
t
i
iA
Exponential Smoothing ExampleDuring the past 8 quarters, the Port of Baltimore has unloaded large quantities of grain. ( = .10). The first quarter forecast was 175..
Quarter Actual1 180
2 1683 1594 1755 190
6 2057 1808 1829 ?
Find the forecast for the 9th quarter.
Exponential Smoothing SolutionFt = Ft-1 + 0.1(At-1 - Ft-1)
QuarterQuarter ActualActualForecast, F t
( αα = = .10.10))
1 180 175.00 (Given)
22 168168 175.00 + .10(180 - 175.00) = 175.50175.00 + .10(180 - 175.00) = 175.50
33 159159 175.50 + .10(168 - 175.50) = 174.75175.50 + .10(168 - 175.50) = 174.75
44 175175
55 190190
66 205205
174.75 + .10(159 - 174.75) = 173.18174.75 + .10(159 - 174.75) = 173.18
173.18 + .10(175 - 173.18) = 173.36173.18 + .10(175 - 173.18) = 173.36
173.36 + .10(190 - 173.36) = 175.02
Exponential Smoothing SolutionFt = Ft-1 + 0.1(At-1 - Ft-1)
Time ActualForecast, F t
(α = .10)
44 175175 174.75 + .10(159 - 174.75) = 173.18174.75 + .10(159 - 174.75) = 173.18
55 190190 173.18 + .10(175 - 173.18) = 173.36173.18 + .10(175 - 173.18) = 173.36
66 205205 173.36 + .10(190 - 173.36) = 175.02173.36 + .10(190 - 173.36) = 175.0277 180180
88
175.02 + .10(205 - 175.02) = 178.02175.02 + .10(205 - 175.02) = 178.02
99 178.22 +178.22 + .10.10(182(182 - 178.22- 178.22)) = = 178.58 178.58
182182 178.02 + .10(180 - 178.02) = 178.22178.02 + .10(180 - 178.02) = 178.22
??
Forecast Effects of Smoothing Constant
Ft = At - 1 + (1- ) At - 2 + (1- )2At - 3 + ...
Weights
Prior Period
2 periods ago
(1 - )
3 periods ago
(1 - )2
=
= 0.10
= 0.90
10% 9% 8.1%
90% 9% 0.9%
= = AAtt-1-1+(1- +(1- ) ) FFtt-1-1
Impact of
0
50
100
150
200
250
1 2 3 4 5 6 7 8 9
Quarter
Actu
al To
nage Actual
Forecast (0.1)
Forecast (0.5)
Choosing
Seek to minimize the Mean Absolute Deviation (MAD)Mean Absolute Deviation (MAD)
If: Forecast error = demand - forecast
Then:n
errorForecast MAD
Using MS-Excel
Naïve MA WMA Exponential Smoothing Method
Associative Model:Regression ( 회기분석 ,回归分析 )
Deviation
Deviation
Deviation
Deviation
Deviation
Deviation
Deviation
Time
Valu
es o
f Dep
ende
nt V
aria
ble
bxaY ˆ
Actual observation
Point on regression line
Actual and the Least Squares Line
0
20
40
60
80
100
120
140
160
1996 1997 1998 1999 2000 2001 2002 2003 2004
Year
Regression Line
Actual Demand
Linear Trend Projection
Used for forecasting linear trend lineAssumes relationship between
response variable, Y, and time, X, is a linear function
Estimated by least squares methodMinimizes sum of squared errors
iY a bX i
Linear Trend Projection Model
Y a bXi i b > 0
b < 0
a
a
Y
Time, X
Least Squares Equations
Equation: ii bxaY
Slope:
xnx
yxnyxb
i
n
i
ii
n
i
Y-Intercept: xbya
X i Y i X i2 Y i
2 X iY i
X 1 Y 1 X 12 Y 1
2 X 1Y 1
X 2 Y 2 X 22 Y 2
2 X 2Y 2
: : : : :
X n Y n X n2 Y n
2 X nY n
ΣX i ΣY i ΣX i2 ΣY i
2 ΣX iY i
Computation Table ( 계산표 )
Using Regression Model
Year Demand 2000 74 2001 79 2002 80 2003 90 2004 105 2005 142 2006 122
The demand for electrical power at N.Y.Edison over the years 2000 – 2006 is given at the left. Find the overall trend.
Calculation for finding a and b
Year Time Period
Power Demand
x2 xy
2000 1 74 1 74
2001 2 79 4 158
2002 3 80 9 240
2003 4 90 16 360
2004 5 105 25 525
2005 6 142 36 852
2006 7 122 49 854
x=28 y=692 x2=140 xy=3,063
The Trend Line Equation
megawatts 151.56 10.54(9) 56.70 2008in Demand
megawatts 141.02 10.54(8) 56.70 2007in Demand
56.70 10.54(4) - 98.86 xb - y a
10.5428
295
(7)(4)140
86)(7)(4)(98.3,063
xnΣx
yxn -Σxy b
98.867
692
n
Σyy 4
7
28
n
Σxx
222
Actual and Trend ForecastElectric Power Demand
60
70
80
90
100
110
120
130
140
150
160
1997 1998 1999 2000 2001 2002 2003 2004 2005
Year
Answers: ‘how strong is the linear relationship between the variables?’
Coefficient of correlation Sample correlation coefficient denoted rValues range from -1 to +1Measures degree of association
Used mainly for understanding
Correlation ( 상관관계 )
Sample Coefficient of Correlation (想关系数 )
n
i
n
iii
n
i
n
iii
n
i
n
i
n
iiiii
yynxxn
yxyxnr
r = 1 r = -1
r = .89 r = 0
Y
XYi = a + b X i^
Y
X
Y
X
Y
XYi = a + b X i^ Yi = a + b X i
^
Yi = a + b X i^
Coefficient of Correlation and Regression Model
r2 = square of correlation coefficient (r), is the percent of the variation in y that is explained by the regression equation
You want to achieve: No pattern or direction in forecast error
Error (or Bias) = (Yi - Yi) = (Actual - Forecast)
Seen in plots of errors over time Smallest forecast error
Mean square error (MSE) Mean absolute deviation (MAD)
Guidelines for Selecting Forecasting Model
^
Time (Years)
ErrorError
00
Desired Pattern
Time (Years)
Error
0
Trend Not Fully Accounted for
Pattern of Forecast Error
You’re a marketing analyst for Hasbro Toys. You’ve forecast sales with a linear model & exponential smoothing. Which model do you use?
Actual Linear Model Exponential Smoothing
Year Sales Forecast Forecast (.9)
2003 1 0.6 1.02004 1 1.3 1.02005 2 2.0 1.92006 2 2.7 2.02007 4 3.4 3.8
Selecting Forecasting Model Example
MSE = Σ Error2 / n = 1.10 / 5 = 0.220MAD = Σ |Error| / n = 2.0 / 5 = 0.400MAPE = 100 Σ|absolute percent errors|/n= 1.20/5 = 0.240
Linear Model Evaluation
Y i
11224
Y i^
0.61.32.02.73.4
Year
20032004200520062007Total
0.4-0.3 0.0-0.7 0.60.0
Error
0.160.090.000.490.361.10
Error2
0.40.30.00.70.62.0
|Error||Error|Actual
0.400.300.000.350.151.20
MSE = Σ Error2 / n = 0.05 / 5 = 0.01MAD = Σ |Error| / n = 0.3 / 5 = 0.06MAPE = 100 Σ |Absolute percent errors|/n = 0.10/5 = 0.02
Exponential Smoothing Model Evaluation
Year
20032004200520062007Total
Y i11224
Y i1.0 0.01.0 0.01.9 0.12.0 0.03.8 0.2
0.3
^ Error
0.000.000.010.000.040.05 0.3
Error2
0.00.00.10.00.2
|Error||Error|Actual
0.000.000.050.000.05
0.10
Exponential Smoothing Model Evaluation ( 어떤 모델이 더 좋은가 ?)
Linear Model:MSE = Σ Error2 / n = 1.10 / 5 = .220MAD = Σ |Error| / n = 2.0 / 5 = .400MAPE = 100 Σ|absolute percent errors|/n= 1.20/5 = 0.240
Exponential Smoothing Model:MSE = Σ Error2 / n = 0.05 / 5 = 0.01
MAD = Σ |Error| / n = 0.3 / 5 = 0.06
MAPE = 100 Σ |Absolute percent errors|/n = 0.10/5 = 0.02
Measures how well the forecast is predicting actual values
Ratio of running sum of forecast errorsrunning sum of forecast errors (RSFE) to mean absolute deviationmean absolute deviation (MAD) = RSFE/MAD Good tracking signal has low values
Should be within upper and lower control limits
Tracking Signal (跟踪信号 )
Tracking Signal Equation
ˆ
1
MAD
yy
MAD
RSFETS
n
iii
MoMo FcstFcst ActAct ErrorError RSFERSFE AbsAbsErrorError
CumCum MADMAD TSTS
11 100100 9090
22 100100 9595
33 100100 115115
44 100100 100100
55 100100 125125
66 100100 140140
|Error||Error|
Tracking Signal Computation
MoMo ForcForc ActAct ErrorError RSFERSFE AbsAbsErrorError
CumCum MADMAD TSTS
11 100100 9090
22 100100 9595
33 100100 115115
44 100100 100100
55 100100 125125
66 100100 140140
-10-10 -10-10 1010 1010 10.010.0 -1-1
-5-5 -15-15 55 1515 7.57.5 -2-2
|Error||Error|
TS = RSFE/MAD = -15/7.5 = -2
TS = RSFE/MAD = -15/7.5 = -2
Tracking Signal Computation
Plot of a Tracking Signal
Time
Lower control limit
Upper control limit
Signal exceeded limit
Tracking signal
Acceptable rangeMAD
+
0
-
The % of Points included within the Control Limits for a range of 1 to 4 MAD
1 03 2 1 2 3 44
±1 MAD
±2 MAD
±3 MAD
±4 MAD
Number of MAD Related STDEV %
±1 0.798 57.048%
±2 1.596 88.946%
±3 2.394 98.334%
±4 3.192 99.895%
Formula Review
Summary What is Forecasting? Types of Forecasting
Qualitative Methods Quantitative Methods
Quantitative Methods Naïve MA Exponential Smoothing Regression
Evaluation Measures MSE, MAD, MAPE
What to do
HW Example 1, 2, 3, 8, 9, 10 Solved Problems 1, 6, 7