[go: up one dir, main page]
Scribd
Upload
703 views46 pages

The Box Jenkins Using Minitab 3

This document describes the process of identifying the best ARIMA model to fit a time series data set. First, autocorrelation (ACF) and partial autocorrelation (PACF) plots were used to check if the original data was stationary. Since it was not stationary, first differencing was performed. Then, ACF and PACF were analyzed again on the differenced data. Based on significant spikes in these plots, four potential ARIMA models (ARIMA(1,1,0), ARIMA(1,1,1), ARIMA(2,1,1), ARIMA(2,1,2)) were identified and estimated. The residuals of each model were then

Uploaded by

sweet
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
703 views46 pages

The Box Jenkins Using Minitab 3

This document describes the process of identifying the best ARIMA model to fit a time series data set. First, autocorrelation (ACF) and partial autocorrelation (PACF) plots were used to check if the original data was stationary. Since it was not stationary, first differencing was performed. Then, ACF and PACF were analyzed again on the differenced data. Based on significant spikes in these plots, four potential ARIMA models (ARIMA(1,1,0), ARIMA(1,1,1), ARIMA(2,1,1), ARIMA(2,1,2)) were identified and estimated. The residuals of each model were then

Uploaded by

sweet
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 46

Prepare data

Insert the data into


Minitab (copy
paste)

Construct trend
analysis

Autocorrelation

Partial Autocorrelation
Coefficient

The series is not stationary


It does not indicate presence of
seasonal effect
No specific action will be made on these
irregularities.

The ACF and PACF were plotted to


collect more evidence on its stationary
condition.
The ACF figure shows the decaying
pattern
There is 2 values of ACF exceed the
significance limit.

Construct trend
analysis

Autocorrelation

Partial Autocorrelation
Coefficient

Perform First Differencing if the original


data is not stationary.
Zt = yt yt-1

The fitted trend line indicate a slight


negative downward trend.

Repeat the same steps to find ACF and


PACF for first differencing zt

The decaying pattern in both ACF and


PACF has disappeared (Figure 21 & 22)
None of the autocorrelation values
exceed the significance limits.
It can be conclude now, the series may
not be necessarily be perfect stationary
due to unexplainable factors inherent in
such data sets.

The process of identifying the suitable


models to be fitted to the data series
involve with the ACF and PACF.
Close scrutiny and careful judgment of the
location and size of the spikes are essential
to determine the number of lags required.
As a way out of this predicament several
possible models will be specified, estimated
and then performed the necessary
validation/diagnostic tests.
The model picked is the one that give the
superior results.

Based on Figure 21 and Figure 22 and


the number of significant spikes, the
following four models have been
identified and estimated using Minitab.
ARIMA(1,1,0)
ARIMA(1,1,1)
ARIMA(2,1,1)
ARIMA(2,1,2)

ARIMA Model: Total Compound payment


Final Estimates of Parameters
Type
Coef SE Coef T
P
AR 1 0.0792 0.1332 0.59 0.555
Constant
61 1091 0.06 0.956
Differencing: 1 regular difference
Number of observations: Original series 59, after
differencing 58
Residuals: SS = 3864081937 (backforecasts
excluded)
MS = 69001463 DF = 56
Modified Box-Pierce (Ljung-Box) Chi-Square statistic
Lag
12 24 36 48
Chi-Square 15.8 25.3 35.1 36.5
DF
10 22 34 46
P-Value 0.105 0.283 0.415 0.841

Repeat the same process for


1. ARIMA(1,1,1)
2. ARIMA(2,1,1)
3. ARIMA(2,1,2)

Find the best fitted model


The characteristics of the best model
1.
The residuals are white noise
2.
No significant autocorrelation
exist
3.
No significant partial
autocorrelation exist
4.
The stationary condition of the
residuals is achieved

Construct Portmanteau Test

Statistics

Model

ARIMA(1,1,0)

ARIMA(1,1,1)

Calculated
Q

15.8

15.8

DF

10

Tabulated Q

18.30

16.91

Decision (5%
sig. level)

Accept H0

Accept H0

Conclusion

The errors
are white
noise

The errors
are white
noise

MSE

69001463

70219280

ARIMA(2,1,1)

ARIMA(2,1,2)

Checking the values of the calculated


Qs and comparing the tabulated values.
The ARIMA(2,1,2) is the best model since
it have smallest Q statistics and MSE.

Apply your data from assignment 1 using


Box-Jenkins model.
Write an analysis and compare the BoxJenkins model with the best model from
assignment 1
Submit before 15 Mei 2016 10 am

You might also like