plrm.gof {PLRModels}

Goodness-of-Fit tests in PLR models

Description

From a sample {(Y_i, X_{i1}, ..., X_{ip}, t_i): i=1,...,n}, this routine tests the null hypotheses H_0: \beta=\beta_0 and H_0: m=m_0, where:

\beta = (\beta_1,...,\beta_p)

is an unknown vector parameter,

m(.)

is a smooth but unknown function and

Y_i= X_{i1}*\beta_1 +...+ X_{ip}*\beta_p + m(t_i) + \epsilon_i.

Fixed equally spaced design is considered for the "nonparametric" explanatory variable, t, and the random errors, \epsilon_i, are allowed to be time series. The test statistic used for testing H0: \beta = \beta_0 derives from the asymptotic normality of an estimator of \beta based on both ordinary least squares and kernel smoothing (this result giving a \chi^2-test). The test statistic used for testing H0: m = m_0 derives from a Cramer-von-Mises-type functional distance between a nonparametric estimator of m and m_0.

Usage

plrm.gof(data = data, beta0 = NULL, m0 = NULL, b.seq = NULL, 
h.seq = NULL, w = NULL, estimator = "NW", kernel = "quadratic", 
time.series = FALSE, Var.Cov.eps = NULL, Tau.eps = NULL, 
b0 = NULL, h0 = NULL, lag.max = 50, p.max = 3, q.max = 3, 
ic = "BIC", num.lb = 10, alpha = 0.05)

Arguments

data

data[,1] contains the values of the response variable, Y;

data[, 2:(p+1)] contains the values of the "linear" explanatory variables

X_1, ... X_p;

data[, p+2] contains the values of the "nonparametric" explanatory variable, t.

beta0

the considered parameter vector in the parametric null hypothesis. If NULL (the default), the zero vector is considered.

m0

the considered function in the nonparametric null hypothesis. If NULL (the default), the zero function is considered.

b.seq

the statistic test for H0: \beta=\beta_0 is performed using each bandwidth in the vector b.seq. If NULL (the default) but h.seq is not NULL, it takes b.seq=h.seq. If both b.seq and h.seq are NULL, 10 equidistant values between zero and a quarter of the range of {t_i} are considered.

h.seq

the statistic test for H0: m=m_0 is performed using each pair of bandwidths (b.seq[j], h.seq[j]). If NULL (the default) but b.seq is not NULL, it takes h.seq=b.seq. If both b.seq and h.seq are NULL, 10 equidistant values between zero and a quarter of the range of {t_i} are considered for both b.seq and h.seq.

w

support interval of the weigth function in the test statistic for H0: m = m_0. If NULL (the default), (q_{0.1}, q_{0.9}) is considered, where q_p denotes the quantile of order p of {t_i}.

estimator

allows us the choice between “NW” (Nadaraya-Watson) or “LLP” (Local Linear Polynomial). The default is “NW”.

kernel

allows us the choice between “gaussian”, “quadratic” (Epanechnikov kernel), “triweight” or “uniform” kernel. The default is “quadratic”.

time.series

it denotes whether the data are independent (FALSE) or if data is a time series (TRUE). The default is FALSE.

Var.Cov.eps

n x n matrix of variances-covariances associated to the random errors of the regression model. If NULL (the default), the function tries to estimate it: it fits an ARMA model (selected according to an information criterium) to the residuals from the fitted regression model and, then, it obtains the var-cov matrix of such ARMA model.

Tau.eps

it contains the sum of autocovariances associated to the random errors of the regression model. If NULL (the default), the function tries to estimate it: it fits an ARMA model (selected according to an information criterium) to the residuals from the fitted regression model and, then, it obtains the sum of the autocovariances of such ARMA model.

b0

if Var.Cov.eps=NULL and/or Tau.eps=NULL, b0 contains the pilot bandwidth for the estimator of \beta used for obtaining the residuals to construct the default for Var.Cov.eps and/or Tau.eps. If NULL (the default) but h0 is not NULL, it takes b0=h0. If both b0 and h0 are NULL, a quarter of the range of {t_i} is considered.

h0

if Var.Cov.eps=NULL and/or Tau.eps=NULL, (b0, h0) contains the pair of pilot bandwidths for the estimator of m used for obtaining the residuals to construct the default for Var.Cov.eps and/or Tau.eps. If NULL (the default) but b0 is not NULL, it takes h0=b0. If both b0 and h0 are NULL, a quarter of the range of {t_i} is considered for both b0 and h0.

lag.max

if Tau.eps=NULL, lag.max contains the maximum delay used to construct the default for Tau.eps. The default is 50.

p.max

if Var.Cov.eps=NULL and/or Tau.eps=NULL, the ARMA model is selected between the models ARMA(p,q) with 0<=p<=p.max and 0<=q<=q.max. The default is 3.

q.max

if Var.Cov.eps=NULL and/or Tau.eps=NULL, the ARMA model is selected between the models ARMA(p,q) with 0<=p<=p.max and 0<=q<=q.max. The default is 3.

ic

if Var.Cov.eps=NULL and/or Tau.eps=NULL, ic contains the information criterion used to suggest the ARMA model. It allows us to choose between: "AIC", "AICC" or "BIC" (the default).

num.lb

if Var.Cov.eps=NULL and/or Tau.eps=NULL, it checks the suitability of the selected ARMA model according to the Ljung-Box test and the t-test. It uses up to num.lb delays in the Ljung-Box test. The default is 10.

alpha

if Var.Cov.eps=NULL and/or Tau.eps=NULL, alpha contains the significance level which the ARMA model is checked. The default is 0.05.

Details

A weight function (specifically, the indicator function 1_{[w[1] , w[2]]}) is introduced in the test statistic for testing H0: m = m_0 to allow elimination (or at least significant reduction) of boundary effects from the estimate of m(t_i).

If Var.Cov.eps=NULL and the routine is not able to suggest an approximation for Var.Cov.eps, it warns the user with a message saying that the model could be not appropriate and then it shows the results. In order to construct Var.Cov.eps, the procedure suggested in Aneiros-Perez and Vieu (2013) can be followed.

If Tau.eps=NULL and the routine is not able to suggest an approximation for Tau.eps, it warns the user with a message saying that the model could be not appropriate and then it shows the results. In order to construct Tau.eps, the procedures suggested in Aneiros-Perez (2008) can be followed.

The implemented procedures generalize those ones in expressions (9) and (10) in Gonzalez-Manteiga and Aneiros-Perez (2003) by allowing some dependence condition in {(X_{i1}, ..., X_{ip}): i=1,...,n} and including a weight function (see above), respectively.

Value

A list with two dataframes:

parametric.test

a dataframe containing the bandwidths, the statistics and the p-values when one tests H0: \beta = \beta_0

nonparametric.test

a dataframe containing the bandwidths b and h, the statistics, the normalised statistics and the p-values when one tests H0: m = m_0

Moreover, if data is a time series and Tau.eps or Var.Cov.eps are not especified:

pv.Box.test

p-values of the Ljung-Box test for the model fitted to the residuals.

pv.t.test

p-values of the t.test for the model fitted to the residuals.

ar.ma

ARMA orders for the model fitted to the residuals.

Author(s)

German Aneiros Perez [e-mail]

Ana Lopez Cheda [e-mail]

References

Aneiros-Perez, G. (2008) Semi-parametric analysis of covariance under dependence conditions within each group. Aust. N. Z. J. Stat. 50, 97-123.

Aneiros-Perez, G., Gonzalez-Manteiga, W. and Vieu, P. (2004) Estimation and testing in a partial linear regression under long-memory dependence. Bernoulli 10, 49-78.

Aneiros-Perez, G. and Vieu, P. (2013) Testing linearity in semi-parametric functional data analysis. Comput. Stat. 28, 413-434.

Gao, J. (1997) Adaptive parametric test in a semiparametric regression model. Comm. Statist. Theory Methods 26, 787-800.

Gonzalez-Manteiga, W. and Aneiros-Perez, G. (2003) Testing in partial linear regression models with dependent errors. J. Nonparametr. Statist. 15, 93-111.

See Also

Other related functions are plrm.est, par.gof and np.gof.

Examples

# EXAMPLE 1: REAL DATA
data(barnacles1)
data <- as.matrix(barnacles1)
data <- diff(data, 12)
data <- cbind(data,1:nrow(data))

plrm.gof(data)
plrm.gof(data, beta0=c(-0.1, 0.35))



# EXAMPLE 2: SIMULATED DATA
## Example 2a: dependent data

set.seed(1234)
# We generate the data
n <- 100
t <- ((1:n)-0.5)/n
beta <- c(0.05, 0.01)
m <- function(t) {0.25*t*(1-t)}
f <- m(t)
f.function <- function(u) {0.25*u*(1-u)}

x <- matrix(rnorm(200,0,1), nrow=n)
sum <- x%*%beta
epsilon <- arima.sim(list(order = c(1,0,0), ar=0.7), sd = 0.01, n = n)
y <-  sum + f + epsilon
data <- cbind(y,x,t)

## Example 2a.1: true null hypotheses
plrm.gof(data, beta0=c(0.05, 0.01), m0=f.function, time.series=TRUE)

## Example 2a.2: false null hypotheses
plrm.gof(data, time.series=TRUE)