13.12 Problems
By adapting the code in Section 13.1, fit a regression model that includes the intercept and a slope, modeling the effect of Wind. What is the mean wind effect you estimate?
Using the results from the linear regression model fit with no burn-in (Section 13.1.1), calculate the ACF of the
beta
time series usingacf()
. Would thinning more be appropriate? How much?Using the fit of the random walk model to the temperature data (Section 13.3), plot the predicted values (states) and 95% CIs.
To see the effect of this increased flexibility in estimating the autocorrelation, make a plot of the predictions from the AR(1) model (Section 13.4 and the RW model (13.3).
Fit the univariate state-space model (Section 13.5) with and without the autoregressive parameter \(\phi\) and compare the estimated process and observation error variances. Recall that AR(1) without the \(\phi\) parameter is a random walk.
Run the examples related to the EFI challenge. Work through the questions associated with each exercise