Two kinds of tests are provided for inference on the coefficients in a fully functional TRP model: integral test and bootstrap test.
sig.test.int.ff(fit, chypo = 0, idx, weight = TRUE, ncut = 2)
sig.test.boots.ff(fit, chypo = 0, idx, nsim = 1000, plot = FALSE)
gof.test.int.ff(fit, cfitList = NULL, idx, weight = TRUE, ncut = 2)
gof.test.boots.ff(fit, cfitList = NULL, idx, nsim = 1000, plot = FALSE)
gof.test.boots.pf(fit1, fit2, nsim, p = NULL, q = 1)Test statistics and their p-values.
a fitted object from tpr
hypothesized value of coefficients
the index of the coefficients to be tested
whether or not use inverse variation weight
the number of cuts of the interval of interest in integral test
a list of fitted object from cst.fit.ff
the number of bootstrap samples in bootstrap test
whether or not plot
fit of H0 model (reduced)
fit of H1 model (full)
the index of the time-varying estimation in fit2
the index of the time-independent estimation in fit1
Jun Yan jun.yan@uconn.edu
Fine, Yan, and Kosorok (2004). Temporal Process Regression. Biometrika.
tpr