Question

In: Statistics and Probability

Using R: 1. Generate AR(1), AR(2), MA(1), MA(2), and ARMA(1,1) processes with different parameter values, and...

Using R:

1. Generate AR(1), AR(2), MA(1), MA(2), and ARMA(1,1) processes with different parameter values, and draw ACF and PACF. Discuss the characteristics of ACF snd PACF for these processes.

2. Generate AR(1) process {X_t}. Compute the first difference Y_t = X_t - X_(t-1). Draw ACF and PACF of {Y_t}. What can you say about this process? Is it again a AR(1) process? What can you say in general?

3.For the AR(2) processes with the following parameters, determine if AR(2) processes are stationary. Without drawing the graphs, what can you say about ACFs.
(a) ϕ1=1.2, ϕ2=−0.2

(b) ϕ1=0.6, ϕ2=0.3

(c) ϕ1=1.2, ϕ2=−0.7

(d) ϕ1=−0.8, ϕ2=−0.7

4. For the process Xt = ϕXt−2+Zt, determine the range of ϕ for which the process is stationary.

Solutions

Expert Solution

solution (1) :

For the model equation (assumed parameter values), we generate a series of 100 observations, plot this series and check their ACF and PACF plots.

a) AR(1)

Model equation: Xt = 0.9 Xt − 1 + Wt

R code :

library(astsa)
x <- arima.sim(model = list(order = c(1, 0, 0), ar = .9), n = 100) # Generate 100 observations from the AR(1) model

plot(x) # Plot the generated data

acf2(x) # Plot the ACF and PACF

Characteristics :

ACF : tails off

PACF : cuts off after lag 1

b) AR(2)

Model equation: Xt =1.5Xt−1 − 0.75 Xt−2 + Wt

R code :

library(astsa)

x <- arima.sim(model = list(order = c(2, 0, 0), ar = c(1.5, -.75)), n = 100)
plot(x)

acf2(x)

Characteristics :

ACF : tails off

PACF : cuts off after lag 2

c) MA(1)

Model equation: Xt = Wt − 0.8Wt−1

R code :

library(astsa)

x <- arima.sim(model = list(order = c(0, 0, 1), ma = -.8), n = 100)
plot(x)
acf2(x)

Characteristics :

ACF : cuts off after lag 1

PACF : tails off

d) MA(2)

Model equation: Xt = Wt − 0.8Wt−1 + 0.5 Wt−2

R code :

library(astsa)
x <- arima.sim(model = list(order = c(0, 0, 2), ma = c(-.8,.5)), n = 100)
plot(x)
acf2(x)

Characteristics :

ACF : cuts off after lag 2

PACF : tails off

d) ARMA(1,1)

Model equation: Xt = .4Xt−1 + Wt - .8Wt−1

R code :

library(astsa)
x <- arima.sim(model = list(order = c(1, 0, 1), ar = 0.4, ma = - 0.8), n = 100)
plot(x)
acf2(x)

Characteristics :

ACF : tails off

PACF : tails off


Related Solutions

Generate n = 100 observations from each of the three models. - ARMA (1,1) - ARMA...
Generate n = 100 observations from each of the three models. - ARMA (1,1) - ARMA (1,0) -ARMA (0,1) 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
How do you distinguish between AR(1) and ARMA(1,1) in the Box-Jenkins Model?
How do you distinguish between AR(1) and ARMA(1,1) in the Box-Jenkins Model?
For the following processes, compute autocovariance function. (a) AR(2) - use the recursive method; (b) ARMA(1,...
For the following processes, compute autocovariance function. (a) AR(2) - use the recursive method; (b) ARMA(1, 1) - you may use the linear representation from lectures; (c) ARMA(1, 2) - derive a linear representation first!
INFERENCE STATISTICS III QUESTION 1 Derive the forecasting formula for ARMA(1,1) model.
INFERENCE STATISTICS III QUESTION 1 Derive the forecasting formula for ARMA(1,1) model.
(a) Show that the lines r 1 (t) = (2,1,−4) + t(−1,1,−1) and r 2 (s)...
(a) Show that the lines r 1 (t) = (2,1,−4) + t(−1,1,−1) and r 2 (s) = (1,0,0) + s(0,1,−2) are skew. (b) The two lines in (a) lie in parallel planes. Find equations for these two planes. Express your answer in the form ax+by+cz +d = 0. [Hint: The two planes will share a normal vector n. How would one find n?] would one find n?]
1. Define how to find different values of r with the correlation coefficient. 2. Suppose the...
1. Define how to find different values of r with the correlation coefficient. 2. Suppose the results of a hypothesis test indicate that we should not reject the null hypothesis. according to that result, where should we place the test stat with respect to a one-tail test to the right? Place the stat on the axis for the picture below, and explain the reasoning. 3.a survey of 225 people indicate that 174 of them prefer Jimmy John's to Subway. Construct...
in R. Generate a random sample of size 700 from a gamma distribution with shape parameter...
in R. Generate a random sample of size 700 from a gamma distribution with shape parameter 8 and scale parameter 0.1. Create a histogram of the sample data. Draw the true parametric density (for the specified gamma distribution) on the histogram. The curve for the density should be red. (Note: The “true parametric density” is the distribution from which the sample values were generated. It is NOT the kernel density that is estimated from the data.)
Consider the following ARMA(1,1) process (1 − 0.3B)Xt = (1 − 0.2B)Zt where{Zi}∼WN(0,σ^2)withE(Z1)=0andVar(Z1)=σ^2 <∞. (i) Discuss...
Consider the following ARMA(1,1) process (1 − 0.3B)Xt = (1 − 0.2B)Zt where{Zi}∼WN(0,σ^2)withE(Z1)=0andVar(Z1)=σ^2 <∞. (i) Discuss if the process has a causal stationary solution. (ii) Find an MA(∞) representation for Xt. (iii) Find the autocorrelation function for the process {Xt}.
1) Using matplotlib, graph the line of slope -2 going through the point (1,1). 2) Instead...
1) Using matplotlib, graph the line of slope -2 going through the point (1,1). 2) Instead of graphing the previous line, plot 5 evenly-spaced points on that line in the interval [0,1] (including 0 and 1).
1. Assume Y is an exponential random variable with rate parameter λ=2. (1) Generate 1000 samples...
1. Assume Y is an exponential random variable with rate parameter λ=2. (1) Generate 1000 samples from this exponential distribution using inverse transform method (2) Compare the histogram of your samples with the true density of Y. Please provide MATLAB programming code below with graphs if possible. Thank you!
ADVERTISEMENT
ADVERTISEMENT
ADVERTISEMENT