Post on 02-Jun-2018
transcript
8/10/2019 3 Session Scm Demand Management 2014 Iimr
1/17
Demand Forecasting
in a Supply Chain.
Professor
Department of Management Studies
Indian Institute of Technology Delhi
Hauz Khas, New Delhi 110 016, India
Phone: +91-11-26596421 (O); 2659-1991(H); (0)-+91-9811033937 (m)Fax: (+91)-(11) 26862620
Email: ravi1@dms.iitd.ac.in, r.s.research@gmail.com,
http://web.iitd.ac.in/~ravi1
eman orecas ng
Forecasting
Predict the next number in the pattern:
a) 3.7, 3.7, 3.7, 3.7, 3.7, ?
. , . , . , . , . ,
c) 5.0, 7.5, 6.0, 4.5, 7.0, 9.5, 8.0, 6.5, ?
8/10/2019 3 Session Scm Demand Management 2014 Iimr
2/17
Forecasting
Predict the next number in the pattern:
a) 3.7, 3.7, 3.7, 3.7, 3.7, 3.7
. , . , . , . , . ,
c) 5.0, 7.5, 6.0, 4.5, 7.0, 9.5, 8.0, 6.5,
.
9.0
BALANCE OF FORECASTING
EFFORTSReference:
Ravi Shankar, IndustrialEngineering & Management
(2010)
Types of forecasting methods
Rely on data andRely on subjective
opinions from one
Qualitative methods Quantitative methods
techniques.or more experts.
8/10/2019 3 Session Scm Demand Management 2014 Iimr
3/17
Qualitative Methods
Grass Roots
Assessment
Executive Judgment
Historical analogy
Market Research
Panel Consensus
Delphi Method
Qualitative
Methods
Qualitative forecasting methods
Executive Judgement: banking on the experience of executives,
who have dealt with similar situations.
Historical Analogy: identifying another similar market.
Market Research: trying to identify customer habits; new product
ideas.
Grass Roots: deriving future demand by asking the person
closest to the customer.
Panel Consensus: deriving future estimations from the synergy
of a panel of experts in the area.
Delphi Method: similar to the panel consensus but with
concealed identities.
Quantitative forecasting methods
Time Series: models that predict future demand based
on past history trends
Causal Relationship: models that use statistical
techniques to establish relationships between variousitems and demand
8/10/2019 3 Session Scm Demand Management 2014 Iimr
4/17
Components of Demand
Average demand for a period of time
Trend Seasonal element
Random variation
Demand
Time
SeasonalPattern
Demand
Time
Cycle
Demand
Time
Average
Product Demand over Time
Demand
Time
Trend
Random
movement
Demand
Time
Trend withSeasonalPattern
11
Time Series Models
Try to predict the future based on past data
Assume that factors influencing the past willcontinue to influence the future
8/10/2019 3 Session Scm Demand Management 2014 Iimr
5/17
Naive Approach
Demand in next period is the same as
demand in most recent periodAugust Sales = 120
Usually not good
Simple Moving Average
Assumes an average is a good estimator of
future behavior
Used when trend is lesser or absent
n
A+...+A+A+A=F 1n-t2-t1-tt1t
+
+
Ft+1 = Forecast for the upcoming period, t+1
n = Number of periods to be averaged
A t = Actual occurrence in period t
Simple Moving Average
Forecast sales for months 4-6 using a 3-period
moving average.
n
A+...+A+A+A=F 1n-t2-t1-tt1t
+
+
SalesMonth (000)
1 4
2 6
3 54 ?5 ?6 ?
8/10/2019 3 Session Scm Demand Management 2014 Iimr
6/17
2a. Simple Moving Average
Sales Moving Average
n
A+...+A+A+A=F 1n-t2-t1-tt1t
+
+
Forecast sales for months 4-6 using a 3-period
moving average.
Month (000) (n=3)1 4 NA
2 6 NA
3 5 NA4 ?5 ?
(4+6+5)/3=5
6 ?
Simple Moving Average
Sales Moving Averageon (n=3)1 4 NA
2 6 NA
3 5 NA4 3
5 ?
5
6 ?
?
Sales Moving Average
2a. Simple Moving AverageSimple Moving Average
on (n=3)1 4 NA
2 6 NA
3 5 NA4 35 ?
5
6 ?
(6+5+3)/3=4.667
8/10/2019 3 Session Scm Demand Management 2014 Iimr
7/17
Simple Moving Average
Sales Moving Averageon (n=3)1 4 NA
2 6 NA
3 5 NA4 35 7
5
6 ?
4.667?
Sales Moving Average
2a. Simple Moving AverageSimple Moving Average
on (n=3)1 4 NA
2 6 NA
3 5 NA4 3
5 7
5
6 ?4.667
(5+3+7)/3=5
Gives more emphasis to recent data
Weighted Moving Average
1n-tn2-t31-t2t11t Aw+...+Aw+Aw+Aw=F ++
decrease for older data
sum to 1.0
8/10/2019 3 Session Scm Demand Management 2014 Iimr
8/17
Weighted Moving Average: 3/6, 2/6, 1/6
Month Weighted
MovingAverage
1 4 NA
1n-tn2-t31-t2t11tAw+...+Aw+Aw+Aw=F
++
Sales
(000)
2 6 NA
3 5 NA4 31/6 = 5.16756 ?
??
Weighted Moving Average: 3/6, 2/6, 1/6
Month Sales(000)
Weighted
Moving
Average
1 4 NA
1n-tn2-t31-t2t11tAw+...+Aw+Aw+Aw=F
++
2 6 NA
3 5 NA4 3 31/6 = 5.1675 76
25/6 = 4.16732/6 = 5.333
Exponential Smoothing
Assumes the most recent observations have
the highest predictive value
gives more weight to recent time periods
= -t+1 t t tet
Ft+1 = Forecast value for time t+1
At = Actual value at time t
= Smoothing constant
8/10/2019 3 Session Scm Demand Management 2014 Iimr
9/17
Exponential Smoothing
Week (i)Demand (Ai)1 820
2 775Given the weekly demand
data what are the ex onential
Ft+1 = Ft + (At - Ft)
3 680
4 655
5 750
6 802
7 798
8 689
9 775
10
smoothing forecasts for
periods 2-10 using =0.10?
Assume F1=D1
Week Demand 0.1 0.6
1 820 820.00 820.00
2 775 820.00 820.00
Ft+1 = Ft + (At - Ft)
=
i Ai Fi
Simple Moving Average
3 680 815.50 793.00
4 655 801.95 725.20
5 750 787.26 683.08
6 802 783.53 723.23
7 798 785.38 770.49
8 689 786.64 787.00
9 775 776.88 728.2010 776.69 756.28
F2 = F1+ (A1F1) =820+0.1(820820)
=820
Week Demand 0.1 0.6
1 820 820.00 820.00
2 775 820.00 820.00
Ft+1 = Ft + (At - Ft)
=
i Ai Fi
Simple Moving Average
3 680 815.50 793.00
4 655 801.95 725.20
5 750 787.26 683.08
6 802 783.53 723.23
7 798 785.38 770.49
8 689 786.64 787.00
9 775 776.88 728.20
10 776.69 756.28
F3 = F2+ (A2F2) =820+.1(775820)
=815.5
8/10/2019 3 Session Scm Demand Management 2014 Iimr
10/17
Week Demand 0.1 0.61 820 820.00 820.00
2 775 820.00 820.00
Ft+1 = Ft + (At - Ft)
=
i Ai Fi
Simple Moving Average
3 680 815.50 793.00
4 655 801.95 725.20
5 750 787.26 683.08
6 802 783.53 723.23
7 798 785.38 770.49
8 689 786.64 787.00
9 775 776.88 728.20
10 776.69 756.28
Week Demand 0.1 0.6
1 820 820.00 820.00
2 775 820.00 820.00
Ft+1 = Ft + (At - Ft)
= =
i Ai Fi
Exponential Smoothing
3 680 815.50 793.00
4 655 801.95 725.20
5 750 787.26 683.08
6 802 783.53 723.23
7 798 785.38 770.49
8 689 786.64 787.00
9 775 776.88 728.2010 776.69 756.28
2.Collecthistoricaldata
1.Identifythepurpose
3.Examinedata(plot)
4.Selectappropriatemodels
5.Computeforecastsforhistorical
o se a orecas ng e o
7b.Adjust
parameters
orselectnew
model
9.Monitorresults
8.Includequalitativeinformation
7a.Forecast overplanninghorizon
6.Isaccuracyacceptable?
Yes
No
30
8/10/2019 3 Session Scm Demand Management 2014 Iimr
11/17
Measures of Forecast Error
MAD =
A - F
n
t tt=1
n
et
a. MAD = Mean Absolute Deviation
b. MSE = Mean Squared Error ( )
n
F-A
=MSE 1=t
2
tt
Ideal values =0 (i.e., no forecasting error)
MSE=RMSEc. RMSE = Root Mean Squared Error
MAD =
A - F
n
t tt=1
n
FtAt
= 40
4=10
MEAN ABSOLUTE DEVIATION
(MAD)
Month Sales Forecast
1 220 n/a
2 250 255
3 210 205
4 300 320
5 325 315
55
20
10
|At Ft|
= 40
What are the Mean Squared Error (MSE) and
Root Mean Squared Error (RMSE) values?
FtAt
= 550
4=137.5
( )
n
F-A
=MSE
n
1=t
2
tt RMSE = 137.5
=11.73
Month Sales Forecast
1 220 n/a
2 250 255
3 210 205
4 300 320
5 325 315
55
20
10
|At Ft| (At Ft)2
2525
400
100
= 550
8/10/2019 3 Session Scm Demand Management 2014 Iimr
12/17
Measuring Bias in Forecast: Tracking signal
The tracking signal is a measure of how often our
estimations have been above or below the actual value. It
is used to decide when to re-evaluate using a model.
RSFETS = =
n
tt )F(ARSFE=
Positive tracking signal: most of the time actualvalues are above our forecasted values
Negative tracking signal: most of the time actualvalues are below our forecasted values
If TS > 4 or < -4, investigate!
Example of Tracking Signal
2/7/2014 35
Linear regression for Forecasting
Linear regression is based on
1. Fitting a straight line to data
2. Explaining the change in one variable through changes
in other variables.
By using linear regression, we try to explore which
independent variables affect the dependent variable
dependent variable = a + b (independent variable)
8/10/2019 3 Session Scm Demand Management 2014 Iimr
13/17
8/10/2019 3 Session Scm Demand Management 2014 Iimr
14/17
What does that mean?
Coke Sales
Average
Monthly
Temperature
Least Squares Method of Linear Regression
Then the line is defined by
bXaY +=
xbya =
22 xnx
yxnxyb
=
Month Advertising Sales X 2 XY
January 3 1 9.00 3.00
Februar 4 2 16.00 8.00
y = a + b X
Regression Example
=
22 xnx
yxnxyb xbya =
March 2 1 4.00 2.00
April 5 3 25.00 15.00
May 4 2 16.00 8.00
June 2 1 4.00 2.00
July
TOTAL 20 10 74 38
8/10/2019 3 Session Scm Demand Management 2014 Iimr
15/17
Is it always possible to use it?
Only if the power supply can be assembled in small lead time
Power supply assembly should be at the end of themanufacturing process
Case Study#1: HP desktop
43
Boardassembly
Hard diskAssembly
TestingPowersupply110 V
Boardassembly
Hard diskassembly
Testing
Powersupply110 V
Powersupply220 V
TestingPowersupply220 V
Delayed product
differentiation
Productpostponement
Case Study#1: HP desktop
Power
su l
Product Product
Month 110 V PC 220 V PC
1 10000 8000
44
Board
assembly
Hard disk
assemblyTesting
110 V
Power
supply
220 V
2 14000 4000
3 16000 2500
4 12000 6500
5 18000 2000
6 15000 4000
7 14000 3000
8 11000 7000
9 13000 5000
10 11000 6000
Forecast accuracy improves at different levels
110 V 220 V Total
Months Demand MA(4 ) Error Demand M A(4) Error Demand MA(4) Error
1 10000 8000 18000
2 14000 4000 18000
3 16000 2500 18500
4 12000 6500 18500
5 18000 13000 -5000 2000 5250 3250 20000 18250 -1750
(10000+14000+16000+12000)/4)
13000-18000
45
6 15000 15000 0 4000 3750 -250 19000 18750 -250
7 14000 15250 1250 3000 3750 750 17000 19000 2000
8 11000 14750 3750 7000 3875 -3125 18000 18625 625
9 13000 14500 1500 5000 4000 -1000 18000 18500 500
10 11000 13250 2250 6000 4750 -1250 17000 18000 1000
MAD 2291.67 1604.17 1020.83
ForecastAccuracy 83.23% 64.35% 94.38%
(5000+1250+3750+1500+2250) / 6100-[(5+1.25+3.75+1.5+2.25)/(18+15+14+11+13+11)]100
8/10/2019 3 Session Scm Demand Management 2014 Iimr
16/17
Case Study#2: Aggregate Forecast
2/7/2014 (c) Dr. Ravi Shankar, AIT
(2008-09)
46
Learning Lesson of Case 1
What are the Learning
Lessons of this case
47
Study?
General Guiding Principles for Forecasting
G 40: Forecasts are more accurate for largergroups of items.
G 41: Forecasts are more accurate for shorterper o s o me.
G 42: Every forecast should include anestimate of error.
G 43: Before applying any forecasting method,the method should be tested and evaluated.
8/10/2019 3 Session Scm Demand Management 2014 Iimr
17/17
Product Redesign Helps Supply Chain
Competitiveness
G 44: Delayed product differentiation is the key tothis redesign
G 45: Similarly, forecast at the most upstream of thesupply chain (if possible)
49
: poss e, never use orecas n orma on athe lower levels. At the lower levels, decisionsshould be based on actual demand