BibliU Print 9781292257914
BibliU Print 9781292257914
7
Demand Forecasting
in a Supply Chain
LEARNING OBJECTIVES
After reading this chapter
chapter,, you will be able to
7.1 Understand the role of forecasting for both 7.4 Analyze demand forecasts to estimate
an enterprise and a supply chain. forecast error
error..
7.2 Identify the components of a demand 7.5 Use Excel to build time-series forecasting
forecast and some basic approaches to models..
models
forecasting.
7.3 Forecast demand using time-series
methodologies given historical demand
data in a supply chain.
A
ll supply chain decisions made before demand has material
materialized
ized are made to a forecast.
In this chapter
chapter,, we explain ho
how
w historical demand information can be used to forecast
future demand and how these forecasts af affect
fect the supply chain. We describe sev
several
eral
methods to forecast demand and estimate a forecast’
forecast’ss accuracy
accuracy.. We then discuss how these
methods can be implemented using Microsoft Excel.
(apush process). Farther up the supply chain, the paint factory that produces the base also needs
forecasts to determine its own production and in inventory
ventory lev
levels.
els. The paint factory’
factory’ss supplie
suppliers
rs also
need forecasts for the same reason. When each stage in the supply chain makes its own separate
forecast, these forecasts are often very di different.
fferent. The result is a mismatch between supply and
demand. When all stages of a supply chain work together to produce a collaborativ collaborativee forecast,
however
howe ver,, the forecast tends to be much more accurate. The resulting forecast accuracy enables
supply chains to be both more responsiv
responsivee and more eff
efficient
icient in serving their customers. Leaders
in many supply chains, from electronics manufacturers to packaged-goods retailers, ha have
ve
improved
improv ed their ability to match supply and demand by moving to toward
ward collaborati
collaborative
ve forecasting.
Consider the value of collaborati
collaborative
ve forecasting for Coca-Cola and its bottlers. Coca-Cola
decides on the timing of promotions based on the demand forecast ov overer the coming quarter
quarter.. Pro
Pro--
motion decisions are then incorporated into an updated demand forecast. The updated forecast is
essential for the bottlers to pl plan
an their capacity and production decisions. A bottler operating
without an updated forecast based on the promotion is unlik unlikely
ely to have sufsufficient
ficient supply avavail
ail--
able for Coca-Cola, thus hurting supply chain prof profits.
its.
Mature products with stable demand, such as milk or paper towels, are usually easiest to
forecast. Forecasting and the accompan
accompanying
ying managerial decisions are extremely di diff
fficult
icult when
either the supply of raw materials or the demand for the finished product is highly unpredictable.
Fashion goods and man many y high-tech products are examples of items that are dif difficult
ficult to forecast.
In both instances, an estimate of forecast error is essential when designing the supply chain and
planning its response.
Before we begin an in-depth discussion of the components of forecasts and forecasting
methods in the supply chain, we brief briefly
ly list characteristics of forecasts that a manager must
understand to design and manage his or her supply chain effecti effectively
vely..
1. Forecasts are al always
ways inaccurate and should thus include both the expected value of the
forecast and a measure of forecast error error.. To understand the importance of forecast error error,,
consider two car dealers. One of them expects sales to range between 100 and 1,900 units,
whereas the other expects sales to range between 900 and 1,100 units. Ev Even
en though both
dealers anticipate avaverage
erage sales of 1,000, the sourcing policies for each dealer should be
very di
different,
fferent, gi
given
ven the dif
difference
ference in forecast accuracy
accuracy.. Thus, the forecast error (or demand
uncertainty) is a kekey
y input into most supply chain decisions. Unfortunately
Unfortunately,, most firms do
not maintain any estimates of forecast error error..
2. Long-term forecasts are usually less accurate than short-term forecasts; that is, long-term
forecasts hav
havee a larger standard dedeviation
viation of error relativ
relativee to the mean than short-term fore
fore--
casts. Forecasting demand a month into the future is harder than forecasting demand a day
into the future. Sev
Seven-Elev
en-Elevenen Japan has exploited this key property to impro improveve its perfor
perfor--
mance. The compan
company y has instituted a replenishment process that enables it to respond to an
order within hours. For example, if a store manager places an order by 10 a.m., the order is
delivered
deli vered by 7 p.m. the same day day.. Therefore, the manager has to forecast what will sell that
night only less than 12 hours before the actual sale. The short lead time allo allows
ws a manager
to take into account current information that could af affect
fect product sales, such as the weather
weather..
This forecast is lik
likely
ely to be more accurate than if the store manager had to forecast demand
a week in adv
advance.
ance.
3. Aggregate forecasts are usually more accurate than disaggre disaggregate
gate forecasts, as they tend to
havee a smaller standard deviation of error relati
hav relative
ve to the mean. For example, it is easy to
forecast the gross domestic product (GDP) of the United States for a giv given
en year with less
than a 2 percent error
error.. Howe
However
ver,, it is much more dif difficult
ficult to forecast yearly rerevenue
venue for a
company with less than a 2 percent error error,, and it is ev
even
en harder to forecast revenue for a
given
giv en product with the same degree of accuracyaccuracy.. The key di difference
fference among the three fore
fore--
casts is the degree of aggre
aggregation.
gation. The GDP is an aggregation across man any
y companies, and
the earnings of a company are an aggre aggregation
gation across several product lines. The greater the
aggregation, the more accurate the forecast.
SUMMARY
SUMMARY OF LEARNING OBJECTIVE 1
Forecasting is a key input for virtually every design and planning decision made in a supply
chain. It is important to recognize that all forecasts are likely to be wrong. Thus, an estima
estima--
tion of forecast error is essential to effectively use the forecast. Reducing the forecast hori
hori--
zon (by reducing the lead time of the associated decision) and aggregation are two
effective approaches to decrease forecast error
error.. A relatively recent phenomenon, however
however,,
is to create collaborative forecasts for an entire supply chain and use these as the basis for
decisions. Collaborative forecasting greatly increases the accuracy of forecasts and allows
the supply chain to maximize its performance. Without collaboration, supply chain stages
farther from demand will likely have poor forecasts that will lead to supply chain ineffiineffi--
ciencies and a lack of responsiveness.
It is important to recognize that past demand is not the same as past sales. Often, compacompa--
nies make the mistake of looking at historical sales and assuming that this is what the historical
demand was. To get true demand, howe howeverver,, adjustments need to be made for unmet demand due
to stockouts, competitor actions, pricing, and promotions. Failure to do so results in forecasts
that do not represent the current reality
reality.. Similarly
Similarly,, a company must account for all factors that are
likely
likely to affect demand before it can select an appropriate forecasting methodology
methodology.. For example,
historically a firm may hahave
ve experienced lo loww demand for chicken noodle soup in July and high
demand in December and January
January.. If the firm decides to discount the product in July
July,, the situa
situa--
tion is lik
likely
ely to change, with some of the future demand shifting to the month of July July.. The firm
should make its forecast taking this factor into consideration.
Forecasting methods are classi
classified
fied according to the follo
following
wing four types:
1. Qualitative: Qual
Qualitativ
itativee forecasting methods are primarily subjecti
subjective
ve and rely on human
judgment. The
Theyy are most appropriate when little historical data are available or when
experts ha
have
ve market intell
intelligence
igence that may affect the forecast. Such methods may also be
necessary to forecast demand sev several
eral years into the future in a new industry
industry..
2. Time series: T Time-series
ime-series forecasting methods use historical demand to make a forecast.
They are based on the assumption that past demand history is a good indicator of future
demand. These methods are most appropri appropriate
ate when the basic demand pattern does not vary
significantly
significantly from one year to the next. These are the simplest methods to implement and
can serve as a good starting point for a demand forecast.
3. Causal: Causal forecasting methods assume that the demand forecast is highly correlated
with certain factors in the en environment
vironment (the state of the economy
economy,, interest rates, etc.).
Causal forecasting methods find this correl correlation
ation between demand and environmental fac fac--
tors and use estimates of what en environmental
vironmental factors will be to forecast future demand. For
example, product pricin
pricing g is strongly correlated with demand. Co Companies
mpanies can thus use
causal methods to determine the impact of price promotions on demand.
4. Simulation: Simulation forecasting methods imitate the consumer choices that gi give
ve rise to
demand to arriv
arrivee at a forecast. Using simulation, a firm can combine time-series and causal
methods to answer such questions as: What will be the impact of a price promotion? What
will be the impact of a competitor opening a store nearby? Airlines simul simulate
ate customer
buying beha
behavior
vior to forecast demand for higher-fare seats when no seats are available at
lower fares.
A company may find it difdifficult
ficult to decide which method is most appropriate for forecast
forecast--
ing. In fact, se
several
veral studies hav
havee indicated that using multiple forecasting methods to create a
combined forecast is more effecti
effective
ve than using an
anyy one method alone.
In this chapter
chapter,, we deal primarily with time-series methods, which are most appropriate
when future demand is related to historical demand, gro growth
wth patterns, and any seasonal patterns.
With any forecasting method, there is always a random element that cannot be explained by his-
torical demand patterns. Therefore, an anyy observed demand can be broken dodown
wn into a systematic
and a random component:
Observed demand (O) = systamatic component (S) + random component (R)
component
The systematic component measures the expected value of demand and consists of what
we will call level, the current deseasonalized demand; trend, the rate of gro growth
wth or decline in
demand for the next period; and seasonality, the predictable seasonal f luctuations in demand.
The random component is the part of the forecast that de deviates
viates from the systematic part. A
company cannot (and should not) forecast the direction of the random component. All a com com--
pany can predict is the random component’
component’ss expected size and varivariability
ability,, which provides a
measure of forecast error
error.. The objecti
objective
ve of forecasting is to filter out the random component
(noise) and esti
estimate
mate the systematic component. The forecast error measures the di difference
fference
between the forecast and actual demand. On av average,
erage, a good forecasting method has an error
whose size is comparable to the random component of demand. A manager should be skeptical
of a forecasting method that claims to ha have
ve no forecasting error on historical demand. In this
case, the method has merged the historical random component with the systematic component.
As a result, the forecasting method will lik
likely
ely perform poorly
poorly..
It is important to understand that forecasting is not just a quantitativ
quantitativee exercise. For effec
effec--
tivee forecasting, an organization must incorporate the follo
tiv following
wing fiv
fivee points into its forecasting
process:
1. Understand the objectiv
objectivee of forecasting.
2. Integrate demand planning and forecasting throughout the supply chain.
3. Identify the major factors that inf
influence
luence the demand forecast.
4. Forecast at the appropriate le
level
vel of aggreg
aggregation.
ation.
5. Establish performance and error measures for the forecast.
this brand will be high in July 2018, because this will occur only if the same brand is promoted
again in July 2018 and other brands resp respond
ond as they did the prev
previous
ious year
year.. When making the
demand forecast, the supermarket must understand what the demand would ha have
ve been in the
absence of promotion acti
activity
vity and how demand is af affected
fected by promotions and competitor actions.
A combination of these pieces of information will allow the supermark
supermarket et to forecast demand for
July 2018, giv
given
en the promotion activity planned for that year
year..
On the supply side, a company must consider the available supply sources to decide on the
accuracy of the forecast desired. If alternate supply sources with short lead times are available, a
highly accurate forecast may not be especially important. Ho However
wever,, if only a single supplier with
a long lead time is av
available,
ailable, an accurate forecast will hav
havee great value.
On the product side, a firm must knoknoww the number of variants of a product being sold and
whether these variants substitute for or complement one another
another.. If demand for a product inf
influlu--
ences or is inf
influenced
luenced by demand for another product, the two forecasts are best made jointly jointly..
For example, when a firm introduces an impro improved
ved version of an existing product, it is likely that
the demand for the existing product will decl decline
ine because customers will buy the improv
improved
ed ver
ver--
sion. Although the decline in demand for the original product is not indicated by historical data,
the historical demand is still useful in that it allo
allows
ws the firm to estimate the combined total
demand for the two versions. Clearly
Clearly,, demand for the two products should be forecast jointly
jointly..
SUMMARY
SUMMARY OF LEARNING OBJECTIVE 2
Demand consists of a systematic and a random component. The systematic component
measures the expected value of demand. The random component measures fluctuations in
demand from the expected value. The systematic component consists of level, trend, and
seasonality. Level measures the current deseasonalized demand
demand.. Trend measures the cur
cur--
rent rate of growth or decline in demand. Seasonality indicates predictable seasonal
Static Methods
A static method assumes that the estimates of le
level,
vel, trend, and seasonal
seasonality
ity within the systematic
component do not vary as ne new
w demand is observed. In this case, we estimate each of these
parameters based on historical data and then use the same values for all future forecasts. In this
section, we discuss a static forecasting method for use when demand has a trend as well as a
seasonal component. We assume that the systematic component of demand is mixed; that is,
Systematic component = (level + trend) * seasonal factor
A similar approach can be appl
applied
ied for other forms as well. We begin with a few basic
definitions:
• L = estimate
estimateof
of lev
level
el at t = 0 (the deseasonalized demand estimate during Period t = 0 )
• T = estimate
estimateof
of trend (increase or decrease in demand per period)
• St = estimate
estimateof
of seasonal factor for Period t
• Dt = actual
actualdemand
demand observed in Period t
• Ft = forecast
forecastof
of demand for Period t
In a static forecasting method, the forecast in Period t for demand in Period t + l is a
product of the le vel in Period t + land the seasonal factor for Period t + l. Let L be the lev
level level
el at
period 0. The le vel in Period t + lis the sum of the lev
level el in Period 0 (L) and (t + l)times the
level
trend T. The forecast in Period t for demand in Period t + lis thus giv
given
en as
Ft + l = [L + (t + l)T]St + l(7.1)
We now describe one method for estimating the three parameters L, T , and S. As an
example, consider the demand for rock salt used primarily to melt sno snow
w. This salt is produced
by a firm called Tahoe Salt, which sells its salt through a variety of independent retailers around
the Lake Tahoe area of the Sierra Ne
Nevvada Mountains. In the past, Tahoe Salt has relied on esti
esti--
mates of demand from a sample of its retailers, but the compan
company y has noticed that these retailers
always overestimate their purchases, leaving Tahoe (and even some retailers) stuck with excess
inventory
inventory.. After meeting with its retailers, Tahoe has decided to produce a collaborative forefore--
cast. Tahoe Salt wants to work with the retailers to create a more accurate forecast based on the
actual demand for their salt. Quarterly retail demand data for the past three years are shown in
Table 7-1 and charted in Figure 7-1.
In Figure 7-1, observe that demand for salt is seasonal, increasing from the second quarter of
a giv
given
en year to the first quarter of the follo
following
wing year
year.. The second quarter of each year has the low
low--
est demand. Each cycle lasts four quarters
quarters,, and the demand pattern repeats every year
year.. There is also
a growth trend in the demand, with sales grogrowing
wing over the past three years. The company estimates
that growth will continue in the coming year at historical rates. We now describe the follofollowing
wing two
steps required to estimate each of the three parameters—lev
parameters—level,el, trend, and seasonal factors.
1. Deseasonalize demand and run linear regression to estimate le
level
vel and trend.
2. Estimate seasonal factors.
ESTIMATING LEVEL AND TREND The objective of this step is to estimate the le
level
vel at Period 0
and the trend. We start by deseasonalizing the demand data. Deseasonalized demand represents
the demand that would ha have
ve been observed in the absence of seasonal f luctuations. The
periodicity (p) is the number of periods after which the seasonal cycle repeats. For Tahoe Salt’
Salt’ss
demand, the pattern repeats ev every
ery year
year.. Given that we are measuring demand on a quarterly
basis, the periodicity for the demand in Table 7-1 is p = 4.
50,000
40,000
30,000
Demand
20,000
10,000
0
1, 2 1, 3 1, 4 2, 1 2, 2 2, 3 2, 4 3, 1 3, 2 3, 3 3, 4 4, 1
Period
Figure 7-1 Quarterly Demand at Tahoe Salt
a
t - 1 + (p/2)
J D t - (p/2) + Dt + (p/2) + 2Di R n (2p) for p even
i = t + 1 - (p/2)
Dt = d (7.2)
a
t + [( p - 1)/2]
D i/p for p odd
i = t - [(p - 1)/2]
In our example, p = 4 is ev
even.
en. For t = 3,
3,we
we obtain the deseasonalized demand using
Equation 7.2 as follows:
With this procedure, we can obtain deseasonalized demand between Periods 3 and 10 as
shown in Figures 7- 2 and 7- 3 (all details are available in the accompan
accompanying
ying spreadsheet
Chapter 7 -T
-Tahoe-salt
ahoe-salt).
).
The following linear relationship exists between the deseasonal
deseasonalized
ized demand,
demand,D
Dt ,and time
timett,
based on the change in demand ov over
er time:
Dt = L + Tt (7.3)
In Equation 7.3,
7.3,D
Dt represents deseasonalized demand and not the actual demand in Period
t, L represents the level or deseasonalized demand at Period 0, and T represents the rate of gro
growth
wth
of deseasonalized demand or tr trend.
end. W
Wee can estimate the values of L and T for the deseasonal
deseasonalized
ized
demand using linear re regression
gression with deseasonalized demand (see Figure 7-2) as the dependent
variable
vari able and time as the independent variable. Such a re regression
gression can be run using Microsoft
Figure 7-
7-2
2Excel Workbook with Deseasonalized Demand for Tahoe Salt
Excel (Data Data Analysis Regression).This sequence of commands opens the Re Regression
gression dia
dia--
log box in Excel. For the Tahoe Salt workbook in Figure 7-2, in the resulting dialog box, we enter
Input Y Range: C4:C11
Input X Range: A4:A11
and click the OK button. A new sheet containing the results of the re regression
gression opens up (see work
work--
sheet Re
Regression-1
gression-1).). This new sheet contains estimates for both the initial lelevel
vel L and the trend T.
The initial le
level,
vel, L, is obtained as the inter
intercept
cept coefficient, and the trend, T, is obtained as the X
variable coeff
coefficient
icient (or the slope) from the sheet containing the regression results. For the Tahoe
Salt example, we obtain L = 18,439 and T = 524 524(all
(all details are av
available
ailable in the worksheet
Regression-1
Regression-1 and numbers are rounded to inte integer
ger values). For this example, deseasonal
deseasonalized
ized
demand Dt for any Period t is thus gigiven
ven by
Dt = 18,439 + 524t (7.4)
Both actual demand and deseasonalized demand are shown in Figure 7-3. It is not appro-
priate to run a linear regression between the original demand data and time to estimate lev
level
el and
trend because the original demand data are not linear and the resulting linear regression will not
be accurate. The demand must be deseasonal
deseasonalized
ized before we run the linear re
regression.
gression.
a Sjp
r-1
jp++i
j=0
Si = (7.6)
r
50,000
30,000
20,000
10,000
0
1 2 3 4 5 6 7 8 9 10
10 11 12
12
Period
Figure 7-3 Deseasonalized Demand for Tahoe Salt
Figure 7-4 Deseasonalized Demand and Seasonal Factors for Tahoe Salt
For the Tahoe Salt example, a total of 12 periods and a periodicity ofp = 4 imply that
there are r = 3 seasonal cycles in the data. We obtain seasonal factors using Equation 7.6 as
Adaptive Forecasting
In adaptiv
adaptivee forecasting, the estimates of lev level,
el, trend, and seasonality are updated after each
demand observation. The main advantage of adapti adaptive
ve forecasting is that estimates incorporate all
new
ne w data that are observed. We now discuss a basic frameframework
work and sev
several
eral methods that can be
used for this type of forecast. The frame
framework
work is provided in the most general setting, when the
systematic component of demand data has the mixed form and contains a le level,
vel, a trend, and a
seasonal factor
factor.. It can easily be modified for the other tw two
o cases, howe
however
ver.. The frame
framework
work can
also be special
specialized
ized for the case in which the systematic component contains no seasonality or
trend. We assume that we havhavee a set of historical data for n periods and that demand is seasonal,
Lt = estimate
estimateof
of lev
level
el at the end of Period t
Tt = estimate
estimateof
of trend at the end of Period t
St = estimate
estimateof
of seasonal factor for Period t
Ft = forecastof demand for Period t (made in Period t - 1or earlier)
forecastof
Dt = actual
actualdemand
demand observed in Period t
Et = Ft - Dt = forecast
forecasterror
error in Period t
adaptivee methods, the forecast for Period t + lin Period t uses the estimate of lev
In adaptiv level
el and
trend in Period t (Lt and Ttrespectiv
respectively)
ely) and is giv
given
en as
Ft + l = (Lt + lT)S t + l (7.7)
The four steps in the adaptiv
adaptivee forecasting framework are as follo
follows:
ws:
Et + 1 = Ft + 1 - Dt + 1 (7.8)
4. Modify estimates: Modify the estimates of le vel (L t + 1), trend (Tt + 1),
level ),and
and seasonal factor
(St + p + 1),
),giv
given
en the error Et + 1 in the forecast. It is desirable that the modifmodification
ication be such
that if the demand is lo lower
wer than forecast, the estimates are revised do downward,
wnward, whereas if
the demand is higher than forecast, the estimates are revised upw upward.
ard.
The revised estimates in Period t + 1 are then used to make a forecast for Period t + 2,
and Steps 2, 3, and 4 are repeated until all historical data up to Period n ha
have
ve been co
covered.
vered. The
estimates at Period n are then used to forecast future demand.
We now discuss various adaptiv
adaptivee forecasting methods. The method that is most appropri
appropriate
ate
depends on the characteristic of demand and the composition of the systematic component of
demand. In each case, we assume the period under consideration to be t.
MOVING AVERAGE The moving average method is used when demand has no observ
observable
able
trend or seasonality
seasonality.. In this case,
Systematic component of demand = level
In this method, the lev
level
el in Period t is estimated as the average demand over the most
recent N periods. This represents an N-period mo
moving
ving average and is evaluated as follo
follows:
ws:
The current forecast for all future periods is the same and is based on the current estimate
of lev
level.
el. The forecast is stated as
Ft + 1 = L t and Ft + n = Lt (7.10)
After observing the demand for Period t + 1,we revise the estimates as follo
1,we follows:
ws:
Lt + 1 = (Dt + 1 + Dt + g + Dt - N + 2)/N, Ft + 2 = L t + 1
To compute the new mo moving
ving average, we simply add the latest observation and drop the old
old--
est one. The re
revised
vised moving average serv
serves
es as the next forecast. The mo
moving
ving average corresponds
to giving the last N periods of data equal weight when forecasting and ignoring all data older than
this ne
new
w moving average. As we increase N, the moving average becomes less responsiresponsive
ve to the
most recently observed demand. We illustrate the use of the moving av average
erage inExample 7-1.
Analysis:
We make the forecast for Period 5 (f
(first
irst week of May) at the end of Period 4 (last week of April).
Thus, assume the current period to be t = 4.
4.Our
Our first objecti
objective
ve is to estimate the le
level
vel in Period
4. Using Equation 7.9 with N = 4,
4,we
we obtain
L4 = (D1 + D2 + D3 + D4 )/4 = (38 + 35 + 77 + 90)/4 = 60
The forecast of demand for Period 5, using Equation 7.10, is expressed as
F5 = L 4 = 60 thousands tons
As the sale in Period 5, D5 ,is 80, we hav
havee a forecast error for Period 5 of
E5 = F 5 - D5 = 60 - 80 = - 20
After observing demand in Period 5, the revised estimate of le
level
vel for Period 5 is gi
given
ven by
L5 = (D2 + D3 + D4 + D5 )/4 = (35 + 77 + 90 + 80)/4 = 70.5
na
1 n
L0 = Di (7.11)
i= 1
The forecast in Period t for all future periods is equal to the current estimate of lev
level
el and is
given
given as
Ft + 1 = L tand F t + n = L t (7.12)
After observing the demand, Dt + 1 , for Period t + 1,we revise the estimate of the le
1,we level
vel as
follows:
Lt + 1 = aDt + 1 + (1 - a)Lt (7.13)
1
Source
Source of data: defra.gov
defra.gov.uk
.uk
n=0
Consider the sales report in Example 7-1, where weekly sales for wheat cereals in Great Britain
has been 38, 35, 77, and 90 thousand tons ov
over
er the four weeks of April 2009. Calculate the sales
forecast for Period 1 (first week of April) using simple exponential smoothing with
ta = 0.1.2
Analysis
In this case we hav
havee demand data for n = 4periods. Using Equation 7.11, the initial estimate of
level
level is expressed by
L0 = a b a Di = 60
1 4
n i =1
Thus, the forecast for Period 1 (using Equation 7.12) is thus giv
given
en by
F1 = L 0 = 60
The observed demand for Period 1 is
isDD1 = 38. The forecast error for Period 1 is gi
given
ven by
E1 = F1 - D1 = 60 - 38 = 22
With a = 0.1,
0.1,the
the revised estimate of le
level
vel for Period 1 using Equation 7.13 is gi
given
ven by
L1 = aD1 + (1 - a)L 0 = (0.1 * 38) + (0.9 * 60) = 57.8
Observe that the estimate of lelevel
vel for Period 1 is lower than for Period 0 because
the demand in Period 1 is lower than the forecast for Period 1. We thus obtain
obtainF
F3 = 55.52,
F4 = 57.67 , and F5 = 60.90
60.90.. Thus, the forecast for Period 5 is 60.90
2
Source
Source of data: defra.gov
defra.gov.uk
.uk
Japan National Tourist Organization has reported a constant increase in number of visitors to
Japan during the last ten years. For example, the number of visitors to Japan from other Asian
countries during the period of 2002–2007 has been 3,417,774; 3,511,513; 4,208,095; 4,627,478;
5,247,125; and 6,130,262 annually
annually.. Forecast the number of visitors for 2008 using trend-
corrected exponential smoothing witha = 0.1, b = 0.2.3
Analysis
The first step is to obtain initi
initial
al estimates of level and trend using linear re
regression
gression.. The estimate
of initial level L0 is obtained as the inter
level cept coefficient and the trend T0 is obtained as variable
intercept
coefficient
coefficient (or the slope). From the given data, we obtain:
L 0 = 2,604,842 and T0 = 548,247
Thus, the forecast for Period 1 (2002) using Equation 7.14 is thus giv
given
en by
F1 = L 0 + T0 = 2,604,842 + 548,247 = 3,153,809
Thus, the observed demand for Period 1 is
isDD1 = 3,417,774.
3,417,774.The
The error for Period 1 is gi
given
ven by
E1 = F1 - D1 = 3,158,089 - 3,417,774 = - 264,685
With a = 0.1, b = 0.2,0.2,the
the revised estimate of le
level
vel and trend for Period 1 using
Equations 7.15 and 7.16 is giv
given
en by
L1 = aD1 + (1 - a)(
)(LL 0 + T0) = (0.1 * 3,417,774) + (0.9 * 3,153,089) = 3,179,558
T1 = b(L1 - L 0 ) + (1 - b)T0 = 0.2 * (3,179,558 - 2,604,842)] + (0.8 * 548,247) = 553,541
Thus,
F2 = L 1 + T1 = 3,179,558 + 553,541 = 3,733,099
3
Data for number of visitors to Japan from overseas quoted from JNTO (Japan National Tourist Organization), www
www..
tourism.jp/english/statistics/inbound.php.