Scribd
Upload
35 views12 pages

Exponential Smoothing Technique

Exponential smoothing technique is used to smooth time series data and forecast future values. It calculates a smoothed value (Ei) as a weighted average of the actual value (Yi) and the previous smoothed value (Ei-1), where the weighting is determined by the smoothing coefficient α between 0 and 1. Autocorrelation refers to the correlation between values of a variable at different times, where the classical linear regression model assumes errors are independent between time periods. The Durbin-Watson test statistically detects the presence of autocorrelation in the residuals (errors) from a regression analysis by calculating a test statistic and comparing it to critical values.

Uploaded by

UTTAM KOIRALA
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
35 views12 pages

Exponential Smoothing Technique

Exponential smoothing technique is used to smooth time series data and forecast future values. It calculates a smoothed value (Ei) as a weighted average of the actual value (Yi) and the previous smoothed value (Ei-1), where the weighting is determined by the smoothing coefficient α between 0 and 1. Autocorrelation refers to the correlation between values of a variable at different times, where the classical linear regression model assumes errors are independent between time periods. The Durbin-Watson test statistically detects the presence of autocorrelation in the residuals (errors) from a regression analysis by calculating a test statistic and comparing it to critical values.

Uploaded by

UTTAM KOIRALA
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 12

Exponential Smoothing Technique

It is used to smooth the data and it is defined as:


For the first period
Ei = Yi
From the second period
Ei = αYi + (1- α) Ei-1
Where

α or w is smoothing coefficient (0 < α<1)

06/21/2021 Prepared by: Pravat Uprety 1


Example 1 (by using α = 0.25)
Month Sales Volume Ei = αYi + (1- α) Ei-1
(Yi) Ei = 0.25 Yi + (1- 0.25) Ei-1
1 74 E1 = 74
2 86 E2 = 0.25 X 86 + 0.75 X 74 = 77
3 45 E3 = 0.25 X 45 + 0.75 X 77 = 69
4 84 E4 = 0.25 X 84 + 0.75 X 69 =72.75
5 36 E5 = 0.25 X 36 + 0.75 X 72.75 = 63.56
6 47 E6 = 0.25 X 47 + 0.75 X 63.56 = 59.42
7 44 E7 = 0.25 X 44 + 0.75 X 59.42 = 55.56

06/21/2021 Prepared by: Pravat Uprety 2


Autocorrelation
One of the assumption of the Classical Linear
Regression Model (CLRM) is that the disturbance
term of the model is independent. Symbolically, it
means that, for the model
Yt= β0 + β1Xt + εt
Cov (εt, εs) = E(εtεs) = 0 for t ≠ s.
This feature of regression disturbance is known
is serial independence or non autocorrelation.
If Cov (εt, εs) = E(εtεs) ≠ 0 for t ≠ s, the we say that
the disturbance term are autocorrelated.

06/21/2021 Prepared by: Pravat Uprety 3


Steps involved in autocorrelation
Procedure of Durbin-Watson test Statistics
The following steps are used for testing the autocorrelation.
Setting up null hypothesis
H 0: ρ = 0
There is no autocorrelation among the error terms
Or
H 0: ρ ≤ 0
There is no positive autocorrelation among the error terms
Or
H 0: ρ ≥ 0
There is no negative autocorrelation among the error terms

Setting up alternative hypothesis


H 1: ρ ≠ 0
There is autocorrelation among the error terms.
Or
H 1: ρ > 0
There is positive autocorrelation among the error terms
Or
H 1: ρ < 0
There is negative autocorrelation among the error terms

06/21/2021 Prepared by: Pravat Uprety 4


Test Statistic
n

 (e  ei i 1 )2
DW  i 2
n

e
i 1
2
i

Decision Criteria
for the value of α, n and k’ we can get the value of dL and dU
from the autocorrelation table

Where

α = level of significance
n = number of periods/years
K’ = no of independent variables.
=1 (if it is not given)

06/21/2021 Prepared by: Pravat Uprety 5


Decision Criteria

06/21/2021 Prepared by: Pravat Uprety 6


Autocorrelation (Example)
Period Error (ei )
1 -5
2 -4
3 -3
4 -2
By using the given information
5 -1
test the presence of
autocorrelation at 5% level of
6 1
significance.
7 2
8 3
9 4
10 5

06/21/2021 Prepared by: Pravat Uprety 7


Solution
Null Hypothesis (H0): ρ = 0
There is no autocorrelation
Alternative Hypothesis (H1): ρ ≠ 0
There is autocorrelation
Test Statistic
n

 i i 1
( e  e ) 2

DW  i 2
n

 i
e 2

i 1

06/21/2021 Prepared by: Pravat Uprety 8


Calculation of test statistic
Period Error ei-1 (ei - ei-1 )2 ei2
(ei )
n

 i i 1
1 -5 - - 25
2 -4 -5 [-4-(-5)]2 = 1 16
( e  e ) 2

3 -3 -4 [-3-(-4)] = 1
2
9 DW  i 2
n
4 -2 -3 1 4  i
e 2

i 1
5 -1 -2 1 1
6 1 -1 4 1
7 2 1 1 4 = 12/110
8 3 2 1 9
9 4 3 1 16
10 5 4 1 25 = 0.109
12 110

06/21/2021 Prepared by: Pravat Uprety 9


Tabulated value
Here n = 10
K’ = 1 (assuming)
α = 0.05 (assuming)
Now from table the value of dL and dU are 0.879 and
1.320 respectively.
Here 0 < DW = 0.109 < dL(0.879)
We reject the Ho
There is autocorrelation among the error terms.
06/21/2021 Prepared by: Pravat Uprety 10
Decision

.109

06/21/2021 Prepared by: Pravat Uprety 11


Thank You

06/21/2021 Prepared by: Pravat Uprety 12

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy