LBTest {PRTest} | R Documentation |
This is an object oriented version which can be used to test an time series for randomness or to test the goodness-of-fit of a fitted time series model.
LBTest(obj, lags = seq(5, 40, 5), SquaredQ = FALSE, BoxPierceQ=FALSE)
obj |
|
lags |
lags to be tested |
NREP |
number of bootstrap replications |
SquaredQ |
use squared residuals |
BoxPierceQ |
if TRUE, the Box-Pierce form of the portmanteau statistic is used |
As shown by Ljung and Box (1978) the statistic
Q_m = n (n+2) sum_{k=1}^m frac{r^2_k}{n-k}
where r_k is the autocorrelation at lag k and n is the length of the time series.
the p-values at the corresponding lags
The parametric bootstrap version of this test is implemented in
our function PRTest
.
A.I. McLeod
Box, G.E.P. and Pierce, D.A. (1970). JASA.
Ljung, G.M. and Box, G.E.P. (1979). The Likelihood Function of Stationary Autoregressive-Moving Average Models. Biometrika 66, 265-270.
The use of squared residuals for detecting nonlinearity and ARCH-like effects is discussed in Li (2004).