In statistics, Exponential Smoothing is a technique that can be applied to

time series data, either to produce smoothed data for presentation, or to

make forecasts. The time series data themselves are a sequence of
observations. The observed phenomenon may be an essentially
random process, or it may be an orderly, but noisy, process. Whereas in
Single Moving Averages the past observations are weighted equally,
Exponential Smoothing assigns exponentially decreasing weights as the
observation get older
Exponential smoothing is commonly applied to financial market and
economic data, but it can be used with any discrete set of repeated
measurements. The raw data sequence is often represented by {xt}, and
the output of the exponential smoothing algorithm is commonly written as
{st} which may be regarded as our best estimate of what the next value of
x will be. When the sequence of observations begins at time t = 0, the
simplest form of exponential smoothing is given by the formulas

                            
where α is the smoothing factor, and 0 < α < 1.
 Meaning
“Exponential smoothing An average method that
exponentially decrease the weighting of old demads.”
It is also called the technique for forcasting .
• Exponential smoothing models are well known and
often used in Operation Management.
• The reasons for their popularity are two :
1) They are readily available in the standard computer
software packages.
2) They require relatively little data storage and
computation .
Some points/advantages related to
Exponential smoothing method
• It tries to overcome the limitations of moving
averages and eliminates the necessity of
keeping extensive records of past data.
• It also tries to screen out the irregularities in the
demand pattern.
• This method allows for trend and takes into
consideration the short term fluctuations in the
determination of the forecast.
• Exponential smoothing is distinguishable by the
special way it weights each past demand.
• The pattern of weights is exponential in form.
• Demand for the most recent period is weighted
most heavily; the weights decrease placed on
successively older periods decrease
exponentially or in other words we can say that
the weight decrease in magnitude the future
back in time the data are weighted ; the
decrease is nonlinear (exponential).
 The Fundamental way of
Exponential Smoothing
• The most recent observation is assigned the highest
weightage.
• It decreases in geometric progression as we move
towards the older ovservations.
• It is considered to be more relastic than moving average
where past observations carries just as much weight as
latest.
• Exponential weighting is, however, a neater and easier
method of achieving a progressive form of weighting.
• Here all one does each period is to suitabily weight and
then add together the current figure and the previous
period’s moving average.
 The Fundamental way of Exponential Smoothing
method is that :

New forecast = α(actual ) + (1- α)(demand)

demand forecast
for most for most
recent recent
period period
Ft = αDt-1 + (1- α) Ft-1
Where
0≤ α ≤ 1, and t is the period.
Ft = the forecast at time t
α = smoothing coefficient
Ft-1 = the forecast at time t-1
Dt-1 = the actual demand at time t-1
Reason for calling it exponential smoothing?
From the above equation :
Ft = αDt-1 + (1- α) Ft-1 ……(1)
Replacing t by t-1 in above equation, then we get
Ft-1 = αDt-2 + (1- α) Ft-2 …….(2)
Simileray
Ft-2 = αDt-3 + (1- α) Ft-3 …….(3)
By putting the value of Ft-1 from eqn (2) in eqn (1) then we get
Ft = αDt-1 + (1- α){ αDt-2 + (1- α) Ft-2}
= αDt-1 + αDt-2 + (1- α) Ft-2 - α² Dt-2 - α (1- α) Ft-2

Ft = αDt-1 + α (1- α) Dt-2 + (1- α)² Ft-2 ….(4)

 Similarly by putting the value of Ft-2 from the
eqn (3) we get :
Ft = αDt-1 + α (1- α) Dt-2 + (1- α)²{αDt-3 + (1- α) Ft-3}
Ft = αDt-1 + α (1- α) Dt-2 + α(1- α)² Dt-3 + (1- α)³ Ft-3…..(5)
The above eqn can be written as :
Ft = α(1- α)ºDt-1 + α(1- α)¹Dt-2 + α(1- α)² Dt-3 + (1- α)³ Ft-3..(6)

We have expended the eqn (1) to obtain eqn (6) . The expansion could
be continued further, but it is not necessary for illustrating our point;
the eqn (6) shows the relative weight that is placed on each past
period’s demand in arriving at a new forecast.
• Since 0 ≤ α ≤ 1 ,the terms α(1- α)º, α(1- α)¹, α(1- α)², (1- α)³ and so
on successively smaller in eqn (6).
• More specifically, these weights decrease exponentially.
• The most recent demand, Dt-1,is given the most weight ,while the
older data are weighted less and less heavily.
NUMERICAL BASED ON THE EXOPNENTIAL SMOOTHING :
QNS : Phoenix General Hospital has experienced irregular, and usually increasing, demand for
disposable kits throughout the hospital. The demand for a disposable plastic tubing in pediatrics
for September was 300 units and for October ,350 units. The old forecast procedure was to use
last year’s average monthly demand as the forecast for each month this year. Last year’s
average monthly demand was 200 units. Using 200 units as the September forecast and a
smoothing coefficient of 0.7 to weight recent demand most heavily, the forecast for this month,
October, would have been (t = October).

Soln :
As we know ,

Ft = αDt-1 + (1- α) Ft-1

Here α = 0.7
Dt-1 = 300
Ft-1 = 200
F october = .7(300) + (1-.7)200
= 210 + 60
F october = 270

The forecast for November would be (t= November)

Ft = αDt-1 + (1- α) Ft-1
= .7(350) + (1-.7)270
F November = 326
 Qns. Forecast the demand for the following series
by exponential smoothing method.

Period
1 2 3 4 5 6 7 8 9 10 11 12
Actual
dem and 10 12 8 11 9 10 15 14 16 15 14 15
 Solution :The forecast for various periods can be calculated in the following tabular form.
Here we consider α =0.1 and then compare the forecast for α =0.7.

Period Actual Ft-1 α=0 et Ft-1 α =0.7 et

Demand Ft Ft

0 10 10
1 10 10 10 0.0 10 10 0.0
2 12 10 10.20 0.2 10 11.40 2.0
3 8 10.20 9.98 -2.20 11.40 9.02 -3.40
4 11 9.98 10.08 1.02 9.02 10.41 1.98
5 9 10.08 9.97 -1.08 10.41 9.42 -1.41
6 10 9.97 9.97 0.03 9.42 9.83 0.58
7 15 9.97 10.47 5.03 9.83 13.45 5.17
8 14 10.47 10.82 3.53 13.45 13.84 0.55
9 16 10.82 11.34 5.18 13.84 15.35 2.16
10 15 11.34 11.71 3.66 15.35 15.11 -0.35
11 14 11.71 11.94 2.29 15.11 14.33 -1.11
12 15 11.94 12.25 3.06 14.33 14.80 0.67
 Now we calculate MAD =E |et| / 12
=2.42 for α = 0.1 and 1.62 for α = 0.7
Since MAD for α = 0.7 is lesser than MAD for α
= 0.1 , α = 0.7 gives better forecast.
 Box- Jenkins Method
• This forecasting approach has been borrowed from the
control theory.
• In a control system, there is a target which has to be
tracked where the target itself is moving or changing its
value or position.
• The control mechanism measures the difference
between the target and actual output.
• Analogous to the target is the demand in the forecasting
system that is sought to be tracked by the forecasting.
• Any error is used alongwith its derivatives and integrals
for improving the forecast for the next period and so on,
The Basic equation is :
New forecast = Old forecast + the direct
term :A(error in old forecast) + time
derivative term : B (change in error
between last time and time immediately
before) + the integral term : C
(sum of the errors so far)
Ft = Ft-1 + (r-1) (et -1 –et-2) +re(et-1) + r1