Scribd
Upload
265 views3 pages

3 9 PDF

This document discusses generating time series data from ARMA(1,1), ARMA(1,0), and ARMA(0,1) models with 100 observations each. It compares the theoretical and sample autocorrelation functions (ACFs) and partial autocorrelation functions (PACFs) of the generated series, finding that in general the theoretical and sample functions match as expected for each model type based on their characteristics.

Uploaded by

Shiva P
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
265 views3 pages

3 9 PDF

This document discusses generating time series data from ARMA(1,1), ARMA(1,0), and ARMA(0,1) models with 100 observations each. It compares the theoretical and sample autocorrelation functions (ACFs) and partial autocorrelation functions (PACFs) of the generated series, finding that in general the theoretical and sample functions match as expected for each model type based on their characteristics.

Uploaded by

Shiva P
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/ 3

Stat5307 HW

November 13, 2017

Problem 3.9:
Generate n = 100 observations from each of the three models discussed in Problem,3.8. Compute the sample
ACF for each model and compare it to the theoretical values. Compute the sample PACF for each of the
generated series and compare the sample ACFs and PACFs with the general results given in Table 3.1.

#Theoretical Values:
#ARMA(1,1)
acf.arma=ARMAacf(ar = .6, ma = .9, lag.max = 20, pacf = FALSE)
pacf.arma=ARMAacf(ar = .6, ma = .9, lag.max = 20, pacf = TRUE)
#ARMA(1,0)
acf.ar=ARMAacf(ar = .6, ma = 0, lag.max = 20, pacf = FALSE)
pacf.ar=ARMAacf(ar = .6, ma = 0, lag.max = 20, pacf = TRUE)
#ARMA(0,1)
acf.ma=ARMAacf(ar = 0, ma = .9, lag.max = 20, pacf = FALSE)
pacf.ma=ARMAacf(ar = 0, ma = .9, lag.max = 20, pacf = TRUE)
par(mfrow=c(2,3))
plot(acf.arma, type ='h', main="ACF of ARMA(1,1)")
abline(h=0)
plot(acf.ar, type ='h', main="ACF of ARMA(1,0) - AR(1)")
abline(h=0)
plot(acf.ma, type ='h', main="ACF of ARMA(0,1)- MA(1)")
abline(h=0)
plot(pacf.arma, type ='h', main="PACF of ARMA(1,1)")
abline(h=0)
plot(pacf.ar, type ='h', main="PACF of ARMA(1,0)- AR(1)")
abline(h=0)
plot(pacf.ma, type ='h', main="PACF of ARMA(0,1) - MA(1)")
abline(h=0)

1
0.8
ACF of ARMA(1,1) ACF of ARMA(1,0) − AR(1) ACF of ARMA(0,1)− MA(1)

0.8

0.8
acf.arma

acf.ma
acf.ar
0.4

0.4

0.4
0.0

0.0

0.0
5 10 15 20 5 10 15 20 5 10 15 20

Index Index Index

PACF of ARMA(1,1) PACF of ARMA(1,0)− AR(1) PACF of ARMA(0,1) − MA(1)


0.8

0.6
0.4
pacf.arma

0.4

0.2
pacf.ma
pacf.ar
0.0

0.2

−0.2
−0.4

0.0

5 10 15 20 5 10 15 20 5 10 15 20

Index Index Index

#Sample Values n= 100


ARMA = arima.sim(n=100, list(order=c(1,0,1), ma=.9, ar=0.6)) #ARMA(1,1)
AR = arima.sim(n=100, list(order=c(1,0,0) , ar=0.6)) # ARMA(1,0) = AR(1)
MA = arima.sim(n=100, list(order=c(0,0,1), ma=.9)) # ARMA(0,1) = MA(1)
par(mfrow=c(2,3))
acf(ARMA, lag.max = 20)
acf(AR, lag.max = 20)
acf(MA, lag.max = 20)
pacf(ARMA, lag.max = 20)
pacf(AR, lag.max = 20)
pacf(MA, lag.max = 20)

2
Series ARMA Series AR Series MA
1.0

1.0

1.0
0.6

0.6

0.6
ACF

ACF

ACF
−0.2 0.2

0.2

0.2
−0.2

−0.2
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20

Lag Lag Lag

Series ARMA Series AR Series MA


0.8

0.4

0.2 0.4
Partial ACF

Partial ACF

Partial ACF
0.4

0.2
0.0

−0.2 0.0

−0.2
−0.6

5 10 15 20 5 10 15 20 5 10 15 20

Lag Lag Lag

Comparision:
ARMA(1,1) model: both the theoretical and sample ACF and PACF values tails off.
AR(1) model: Theoretical ACF tails off and PACF cuts off after lag 1, sample ACF tails off and sample
PACF cuts off after lag 1.
MA(1) model: Theoretical ACF cuts off and PACF tails off, sample ACF cuts off after lag 1 and sample
PACF tails off.
In general, both theoretical and sample ACF and PACF’s are similar for all models, but as the sample size
get larger the sample ACF and PACF should become more similar to theoretical values.

You might also like