Short Term Forecasting – Winters

Demand Management
Chapter 4
Processing,
Influencing, &
Anticipating
Demand
Make
Store
Sell
Buy Move
Buy Move Make
Sell Buy Make
Sell
Move
Move
Store
McGraw-Hill/Irwin
Copyright © 2008 by The McGraw-Hill Companies, Inc. All rights reserved.
Managing the sell side of a business
Supply-Demand Management
"Make, Move, Store"
Plant
Plant
Plant
Warehouse
Customer
Relationship
Management
"Sell"
Customers
Suppliers
Supplier
Relationship
Management
"Buy"
4-2
Key questions
1.
2.
What is the scope of demand management?
What does order processing involve; why is it an important area for management
attention?
3.
4.
What is customer profit potential, & how is it relevant for influencing demand?
What are 5 alternatives for improving forecast accuracy, what do they mean, & how
can they be applied?
How do the tactics of part standardization & postponement of form or place help
improve forecast accuracy?
What is the difference between long term & short term forecasting?
What are 4 long term forecasting methods; what are the risks of
salesperson/customer input?
What are the components of demand, & which component is not forecasted?
How do the moving average, Winters, & focus forecasting methods work?
What is the role of the number of periods in the moving average method, & the
smoothing parameters in the Winters method?
5.
6.
7.
8.
9.
10.
11.
12.
13.
What is the purpose of filtering, & why is it important for computer-based
forecasting?
What do the following principles of nature mean & how are they relevant for
demand management? (1) law of large numbers, (2) trumpet of doom, (3) recency
effect, (4) hockey stick effect, (5) Pareto phenomenon
What are the managerial insights from the chapter?
4-3
Road map
• Processing Demand
• Influencing Demand
• How to Improve Forecast Accuracy
• Long Term Forecasting
• Short Term Forecasting
• Summary
4-4
Scope of demand management
• So what is demand management?
 Concerned with processing, influencing,
and anticipating demand
• We’ll begin with processing demand or, in
more common terms, order processing
or order fulfillment
4-5
Processing Demand
Order processing
• Order processing is usually viewed to span
order booking to order shipment
• Example steps?
 Customer validation, order entry, credit checking,
pricing, design changes, availability checks, delivery
time estimation, notification of shipment, notification
of delays
4-6
Processing Demand
CUSTOMER
RETURNS
ORDER ENTRY AND
CHECKING
Customer Validation
Credit Control Operations…
ER
P
ORDER
INTERRUPTION
ORDER
PICKING AND
ASSEMBLY
CUSTOMER SERVICE
SHIPPING
INVOICING
4-7
Processing Demand
Characteristics
• Can be a complex & time consuming process
dealing largely with information flow
 Susceptible to ad hoc modifications over time in
response to problems (e.g., extra credit check added
due to expensive nonpaying customer a few years
ago)
• A major customer contact point with
organization
 Can significantly impact customer perceptions
• IT advances & high customer impact
 A potential profitable target for improvement
4-8
Processing Demand
Example 1
Benetton
• Electronic loop linking sales agent, factory, & warehouse
• If not available, measurements transferred to knitting
machine for production
• Benetton uses a single warehouse
 Staffed by 8 people & about 230,000 pieces shipped daily
4-9
Processing Demand
Example 2
K-Mart and MasterLock
• Policy for mistake in shipment or invoice
 Strike 1: $10,000, Strike 2: $50,000, Strike 3: lose
business
• MasterLock revamped their order processing function
4-10
Processing Demand
Example 3 – customer tools
• Amazon online order tracking
4-11
Processing Demand
Example 4 – customer tools
• UPS online order tracking
4-12
Processing Demand
Example 4 – continued
• UPS online tools
4-13
Road map
• Processing Demand
• Influencing Demand
• How to Improve Forecast Accuracy
• Long Term Forecasting
• Short Term Forecasting
• Summary
4-14
Influencing Demand
Measure customer profit potential
A simple idea
• Some customers are more profitable than others
• Advancing technologies  more practical to estimate profit
potential of individual customers
• Can guide efforts/investments for customer retention &
acquisition . . . investments to influence demand
• E.g.,
 Electronics manufacturer: reviews historical customer profit before
sending service contract renewal
 Wireless phone firm: churn scores & lifetime value estimates
influence # of customer contacts & attractiveness of offerings
• Ongoing development of data mining methods
4-15
Road map
• Processing Demand
• Influencing Demand
• How to Improve Forecast Accuracy
• Long Term Forecasting
• Short Term Forecasting
• Summary
4-16
Forecasting Alternatives
Motivating example 1
Sunbeam
Improved forecasting led to 45% reduction in
inventory
 Included estimates from top 200 customers
4-17
Forecasting Alternatives
Motivating example 2
Apple
A history of problems forecasting demand
Many components sourced from 1 supplier accurate forecasts are critical
Over $1 billion in unfilled orders during the
crucial holiday season. The CEO (Spindler)
ousted a few months later
4-18
Forecasting Alternatives
Motivating example 3
IBM
Badly misjudged demand in PC business in 1996
– went from being profitable in 1995 to a $200
million loss through 1st half of 1996
4-19
Forecasting Alternatives
Motivating example 4
Christmas 1999 & e-commerce takes
off
Large unanticipated increase in Internet orders –
didn’t ship on time
E.g., Many Toys ‘R Us Christmas orders not delivered
until March – “I will never buy online again”
4-20
Forecasting Alternatives
Improvement alternatives
• Change the forecasting method
 Collect more or different data
 Analyze the information differently
• E.g., involve more people, new forecasting software, spend more
time manually reviewing, focus groups etc.
• Change operations or operating policies
 Introduce early warning mechanisms
 Take advantage of the law of large numbers
 Reduce information delays & leadtimes (trumpet of doom)
 Reduce demand volatility
4-21
Forecasting Alternatives
Early warning
• Change policies so that some (or more)
customers provide earlier commitment of
future demand, e.g.,
 Early bird program for builder markets – discount for
60-day advance order
 Invite large buyers to Aspen in February to view next
year’s skiwear line, & encourage orders
• “Commitment”  asking customers how much
they are likely to buy next quarter
4-22
Forecasting Alternatives
Law of large numbers
Principle of Nature
• As volume increases, relative variability decreases
 Postponement in form or place, e.g.,
• Dell – configure your own PC
• From full product line at 12 regional DCs to full product line at a
single super DC, with 10% of product line stocked at 11 regional
DCs (i.e., fast movers that account for 70% of sales)
 Part standardization, e.g.,
• Arby’s sandwich wrappers; plastic lids with push down drink
indicator
• Intel Pentium processors all the same size
- 2.8 GHz tests out below 2.8 spec can be sold as a 2.66 GHz chip (“downbinning”)
4-23
Forecasting Alternatives
Trumpet of doom
Forecast Error Range over Time
Principle of Nature
• As forecast horizon
increases, accuracy
decreases, e.g.,
Percentage
Forecast 0
Error
0
Time Until Forecast Event
 Reduce production & delivery leadtimes
• Dell pick-to-light system for assembly
 Reduce information delays
• EDI transmission of daily consumer demand up
through multiple levels in the supply chain
4-24
Forecasting Alternatives
Reduce demand volatility
2 Principles of Nature
•
Beware of product proliferation

Pareto analysis – separating the important few from the trivial many

Periodic length of line analysis to critically assess whether to continually
offer “slow movers”

Principle of Nature: Pareto phenomenon – the lion’s share of an aggregate
measure is determined by relatively few factors
• E.g., “the 80-20 rule” – 80% of demand is due to 20% of product line
•
Beware of perverse cycle of promotions – customers wait for sale
before buying, thereby forcing a sale

•
A step further – dynamic pricing to stabilize demand & align with supply
Reduce the hockey stick effect…
4-25
Forecasting Alternatives
Hockey stick effect
Principle of Nature
• Volume tends to pick up towards the end
of a reporting period . . . why?
Jan
Feb
• Look for ways to lessen the effect –
contributes to demand volatility,
inefficiency, poor service
4-26
Forecasting Alternatives
Channel stuffing
One contributor to the hockey stick effect
Lots of sales booked near the end of a quarter,
then sales drop off at the start of the next
quarter
E.g.,
 A large brewer offered a vacation to the salesperson
in each region who sold the most beer to stores over
a 3 month period
 One winner was able to convince a few stores to free
up backroom space and fill it entirely with beer
4-27
Forecasting Alternatives
Improvement alternatives
• We’re about to focus on
methods for predicting
demand
•
short pork bellies
But, important to remember . . . many creative ways to
improve forecast accuracy that have nothing to do with
method
– E.g., early warning incentives, law of large numbers, trumpet of
doom, reduce demand volatility
4-28
Road map
• Processing Demand
• Influencing Demand
• How to Improve Forecast Accuracy
• Long Term Forecasting
• Short Term Forecasting
• Summary
4-29
Long Term Forecasting
Characteristics of long term forecasts
• Single or multi-year horizon
• Monthly or annual time bucket
• Aggregate units
 Input to “long term” decisions
• Accuracy generally more important than short term
forecasts . . . why?
• Tend to use expensive & time consuming methods . . .
due to the preceding point & due to a PON . . . which is?
4-30
Long Term Forecasting
Some alternative methods
• Judgment
• Salesperson & customer input
 Great information source, but beware of bias potential
& recency effect = humans tend to be overly
influenced by recent events
• Outside services
• Causal methods . . . examples?
4-31
Long Term Forecasting
Recency effect
Principle of Nature
Humans tend to overreact to (or be overly
influenced by) recent events
E.g.,
Hughes Electronics Corp. developed an artificial
intelligence based financial trading system. The
developers did this by encoding the wisdom of
Christine Downton, a successful portfolio manager.
One motivation for creating the system is that it is
immune to the recency effect, i.e., humans tend to
get overly fixated on the most recent information.
4-32
Road map
•
Processing Demand
•
Influencing Demand
•
How to Improve Forecast Accuracy
•
Long Term Forecasting
•
Short Term Forecasting
•

Characteristics

Components of demand

Moving average

Winters method

Focus forecasting

Filtering
Summary
4-33
Short Term Forecasting
Long term/short term characteristics
Long term forecasts
Short term forecasts
 Single or multi-year horizon
 Weekly or monthly horizon
 Monthly or annual time bucket
 Daily & weekly time bucket
 Aggregate units (e.g., product/
service categories)
 Detailed units (e.g., SKU)
 Input to “long term” decisions
 Expensive & time consuming
methods
• Accuracy importance
 Input to “short term” decisions
 Inexpensive & quick methods
• Accuracy importance
• Trumpet of doom
• Trumpet of doom
Could argue using 2 different principles of nature that it’s [easier?/harder?] to be
accurate with short term forecasting than with long term forecasting
4-34
Key questions
1.
2.
What is the scope of demand management?
What does order processing involve; why is it an important area for management
attention?
3.
4.
What is customer profit potential, & how is it relevant for influencing demand?
What are five alternatives for improving forecast accuracy, what do they mean, &
how can they be applied?
How do the tactics of part standardization & postponement of form or place help
improve forecast accuracy?
What is the difference between long term & short term forecasting?
What are four long term forecasting methods; what are the risks of
salesperson/customer input?
What are the components of demand, & which component is not forecasted?
How do the moving average, Winters, & focus forecasting methods work?
What is the role of the number of periods in the moving average method, & the
smoothing parameters in the Winters method?
What is the purpose of filtering, & why is it important for computer-based
forecasting?
What do the following principles of nature mean & how are they relevant for
demand management? (1) law of large numbers, (2) trumpet of doom, (3) recency
effect, (4) hockey stick effect, (5) Pareto phenomenon
What are the managerial insights from the chapter?
5.
6.
7.
8.
9.
10.
11.
12.
13.
4-35
Road map
•
Processing Demand
•
Influencing Demand
•
How to Improve Forecast Accuracy
•
Long Term Forecasting
•
Short Term Forecasting
•

Characteristics

Components of demand

Moving average

Winters method

Focus forecasting

Filtering
Summary
4-36
Short Term Forecasting
Example setting...
• Need daily demand forecasts for next 28 days for
thousands of items, as well as forecast accuracy
estimates
 For routine operating decisions, especially relating to
logistics
• Forecasts are updated every week on a rolling basis
• Daily shipment data recorded in database
 Points to time series methods that can be implemented
on a computer
4-37
Short Term Forecasting
Components of demand
1. Underlying average
2. Trend
3. Seasonality (or patterns)
4. “Randomness” (or white noise)
•
Challenge is to estimate the value of components 1 - 3 from
the past, each of which may continually change over time
•
Combine components to develop forecasts for each period in
future

Casting forward the past to predict the future
4-38
Short Term Forecasting – Moving Average
A simple m-period moving average
• Suppose no trend or seasonality – just trying to
estimate the mean
• One option is an m-period moving average:
xt = demand in period t
mat = moving average calculated at end of period t
m = number of periods in moving average calculation
Ft,j = forecast calculated at end of period t for period t+j
mat =(xt-m+1+xt-m+2+...+xt-1+xt)/m
= average during most recent m periods
Ft,j = mat
4-39
Short Term Forecasting – Moving Average
Example
Period (t):
Shipment history:
7
10
8
20
9
25
10
30
11
20
12
40
Forecasts for periods 13, 14, 15, . . . at end of t =12
m=2: (20 + 40)/2 = 30
m=3: (30 + 20 + 40)/3 = 30
m=4: (25 + 30 + 20 + 40)/4 = 29
m=6: (10 + 20 + 25 + 30 + 20 + 40)/6 = 26
4-40
Short Term Forecasting – Moving Average
Why a “moving” average?
Time bucket = 1 week; forecast horizon = 4 weeks
4-period moving average as it “moves” through time
Week of:
As of 1/30
Actual:
Forecast:
1/3
212
1/10
1/17
1/24
167
195
235
1/31
202
As of 2/6
Actual:
Forecast:
212
As of 2/13
Actual:
Forecast:
212
As of 2/20
Actual:
Forecast:
212
167
167
167
195
195
195
235
235
235
2/7
2/14
2/21
2/28
3/7
202
202
202
225
225
225
225
229
229
229
229
235
235
235
3/14
301
301
301
184
184
220
235
4-41
Short Term Forecasting – Moving Average
What about the # of periods in the average?
• As m increases, the method becomes
[more?/less?] responsive to changes in demand
• Selection of m – balancing possibility of
overreacting to changes in the market with not
picking up on real shifts in the mean demand
• Examples of when small or large m might be
appropriate?
4-42
Short Term Forecasting – Moving Average
Some pros/cons
1. Simple (+)
2. Equal weights of history (-)
3. History cut-off beyond m periods (-)
4-43
Road map
•
Processing Demand
•
Influencing Demand
•
How to Improve Forecast Accuracy
•
Long Term Forecasting
•
Short Term Forecasting
•

Characteristics

Components of demand

Moving average

Winters method

Focus forecasting

Filtering
Summary
4-44
Short Term Forecasting – Winters
Old man winters
Winters method used to forecast one period into the future
See how method detects patterns & adapts to market changes over
time
Old Man Winters in Action
600.00
Volum e
500.00
400.00
Actual
300.00
Forecast
200.00
100.00
0.00
0
20
40
60
80
100
Tim e
4-45
Short Term Forecasting – Winters
Key to Winters method
• Winters is an exponential smoothing
method
• Smoothing is based on a key idea
 For each component (which are?), a portion
of difference between estimate & actual is
due to randomness & certain portion due
to real change
4-46
Short Term Forecasting – Winters
Smoothing in action...
• New estimate = old estimate + (some
percentage)(error)
• Smoothes out peaks & valleys (i.e.,
randomness) of actual
4-47
Short Term Forecasting – Winters
Temporarily ignoring trend & seasonality...
• Only estimating the mean component (a.k.a. basic
exponential smoothing)
• Details:
xt = actual in period t
st = smoothed estimate of the mean calculated at the end of
period t
 = smoothing parameter,   [0,1]
st = old estimate + (error) = st-1 + (xt-st-1) = xt + (1- )st-1
Ft,j = forecast calculated at the end of period t for period t+j
Ft,j = st
4-48
Short Term Forecasting – Winters
Example
Time bucket = 1 week; forecast horizon = 4 weeks, s1/30 = 202,  = 0.4
Smoothed estimate of the mean as it is updated over time
Week of:
As of 1/30
Actual:
Forecast:
1/3
212
1/10
1/17
1/24
167
195
235
1/31
202
As of 2/6
Actual:
Forecast:
212
As of 2/13
Actual:
Forecast:
212
As of 2/20
Actual:
Forecast:
212
167
167
167
195
195
195
235
235
235
2/7
2/14
2/21
2/28
3/7
202
202
202
242
242
242
242
219
219
219
219
219
219
219
3/14
301
301
301
184
184
220
219
4-49
Short Term Forecasting – Winters
Visualization of concept
• = forecast
x = actual
x
x
x
x
x
x
x
x
x
Time
4-50
Short Term Forecasting – Winters
Selection of 
• New estimate = old estimate + (error)
• As  increases, does the method become more
or less responsive to changes in the market?
• In practice, usually   [0.1, 0.3]
 Tends to be larger for newer products/services,
smaller for more mature products/services
 Most software packages include capabilities to
recommend a value of  based on history
4-51
Short Term Forecasting – Winters
Adding the seasonality component
Seasonal index = demand in season  mean
demand
x
x = actual
x
x
x
x
x
x
x
Variation in seasonal index
due to randomness, real
change in season impact, or
shift to new overall average?
x
Time
Smoothing used to estimate & update seasonal indices
4-52
Short Term Forecasting – Winters
Method mechanics
st =
smoothed estimate of mean demand in period t calculated at
the end of period t
Mt =
smoothed estimate of index for season corresponding to period
t calculated at the end of period t: season meanoverall mean
,=
smoothing parameters for st, Mt where   [0,1],   [0,1]
st =
=
old estimate + (error) = st-1 + [(xt/Mt-L) - st-1]
(xt/Mt-L) + (1- )st-1
Mt =
=
old estimate + (error) = Mt-L + [(xt/st) - Mt-L]
(xt/st) + (1- )Mt-L
Ft,j =
st  (appropriate index for period t+j) = stMt-L+j
4-53
Short Term Forecasting – Winters
Adaptability in action
L = 2,  = 0.1, b = 0.2,  = 0.3, s0 = 10.0, b0 = 2.0, MOdd = 1.20, MEven = 0.80
F0,1 = [s0 + (1)b0]MO = [10.0 + 2.0](1.20) = 14.4
x2 = 13
F0,2 = [s0 + (2)b0]ME = [10.0 + 4.0](.80) = 11.2
s2 = .1(x2/ME) + .9(s1+b1)
F0,3 = [s0 + (3)b0]MO = [10.0 + 6.0](1.20) = 19.2
= .1(13/.80) + .9(14.4) = 14.6
b2 = .2(s2-s1) + .8(b1)
x1 = 18
= .2(14.6-12.3) + .8(2.1) = 2.1
s1 = .1(x1/MO) + .9(s0+b0) = .1(18/1.20) + .9(12)
= 12.3
ME(new) = .3(x2/s2) + .7[ME(old)]
= .3(13/14.6) + .7(.80) = .83
b1 = .2(s1-s0) + .8(b0) = .2(12.3-10.0) + .8(2.0)
= 2.1
MO(new) = .3(x1/s1) + .7[MO(old)]
= .3(18/12.3) + .7(1.20) = 1.28
F2,1 = [s2 + (1)b2]MO = [14.6 + 2.1](1.28) = 21.4
F2,2 = [s2 + (2)b2]ME = [14.6 + 4.2](.83) = 15.6
F2,3 = [s2 + (3)b2]MO = [14.6 + 6.3](1.28) = 26.8
F1,1 = [s1 + (1)b1]ME = [12.3 + 2.1](.80) = 11.5
F1,2 = [s1 + (2)b1]MO = [12.3 + 4.2](1.28) = 21.1
F1,3 = [s1 + (3)b1]ME = [12.3 + 6.3](.80) = 14.9
4-54
Road map
•
Processing Demand
•
Influencing Demand
•
How to Improve Forecast Accuracy
•
Long Term Forecasting
•
Short Term Forecasting
•

Characteristics

Components of demand

Moving average

Winters method

Focus forecasting

Filtering
Summary
4-55
Short Term Forecasting – Focus
Bernie’s insight…
…or what is focus forecasting?
• An intuitive & successful idea
• Regularly use a # of different methods to
generate forecasts
• Maintain historical accuracy information on each
method
• Use the most accurate method to generate
“official” forecasts
4-56
Short Term Forecasting – Focus
Advertisement
appearing in
APICS The
Performance
Advantage
4-57
Road map
•
Processing Demand
•
Influencing Demand
•
How to Improve Forecast Accuracy
•
Long Term Forecasting
•
Short Term Forecasting
•

Characteristics

Components of demand

Moving average

Winters method

Focus forecasting

Filtering
Summary
4-58
Short Term Forecasting – Filtering
Two types of filters
•
An important feature of computer-based forecasting
systems

Large amounts of data – impractical to manually review all
1. For data input errors (e.g., typos, scanner errors)

If |“actual” - forecast| > limit, then report
2. For unacceptable forecast errors (e.g., warranting
management attention)

If average absolute error > limit, then report

If average error (i.e., bias) > limit, then report
4-59
Road map
• Processing Demand
• Influencing Demand
• How to Improve Forecast Accuracy
• Long Term Forecasting
• Short Term Forecasting
• Summary
4-60
Managerial insights
• Pay attention to order
processing
 Point of high customer
contact
 Makes use of IT – a
rapidly advancing area
• Consider changing your
operations rather than
collecting more data or using
different forecasting
techniques
 (1) early warning, (2) law
of large numbers, (3)
trumpet of doom, (4)
reduce demand volatility
• A simple & pragmatic method
 Old Man Winters is a
practical, powerful, & easy
to implement tool for
casting forward the past to
predict the future
 Can be part of a focus
forecasting approach
4-61
Back to key questions
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
What is the scope of demand management?
What does order processing involve; why is it an important area for management
attention?
What is customer profit potential, & how is it relevant for influencing demand?
What are 5 alternatives for improving forecast accuracy, what do they mean, & how
can they be applied?
How do the tactics of part standardization & postponement of form or place help
improve forecast accuracy?
What is the difference between long term & short term forecasting?
What are 4 long term forecasting methods; what are the risks of
salesperson/customer input?
What are the components of demand, & which component is not forecasted?
How do the moving average, Winters, & focus forecasting methods work?
What is the role of the number of periods in the moving average method, & the
smoothing parameters in the Winters method?
What is the purpose of filtering, & why is it important for computer-based
forecasting?
What do the following principles of nature mean & how are they relevant for
demand management? (1) law of large numbers, (2) trumpet of doom, (3) recency
effect, (4) hockey stick effect, (5) Pareto phenomenon
What are the managerial insights from the chapter?
4-62