summary: Summarizes Fit Information for bigsplines Model

In bigsplines: Smoothing Splines for Large Samples

Description

This function summarizes basic model fit information for a fit bigsplines model.

Usage

## S3 method for class 'bigspline'
summary(object, fitresid = TRUE, chunksize = 10000, ...)
## S3 method for class 'bigssa'
summary(object, fitresid = TRUE, chunksize = 10000, diagnostics = FALSE,...)
## S3 method for class 'bigssg'
summary(object, fitresid = TRUE, chunksize = 10000, diagnostics = FALSE,...)
## S3 method for class 'bigssp'
summary(object, fitresid = TRUE, chunksize = 10000, diagnostics = FALSE,...)
## S3 method for class 'bigtps'
summary(object, fitresid = TRUE, chunksize = 10000, ...)

Arguments

object	Object of class "bigspline" (output from bigspline), class "bigssa" (output from bigssa), class "bigssg" (output from bigssg), class "bigssp" (output from bigssp), or class "bigtps" (output from bigtps).
fitresid	Logical indicating whether the fitted values and residuals should be calculated for all data points in input object.
chunksize	If fitresid=TRUE, fitted values are calculated in chunks of size chunksize.
diagnostics	If diagnostics=TRUE, cosine diagnostics are calculated for each term in the model. These give an approximate break-down of the model R-squared into that accounted for by each term in the model.
...	Ignored.

Details

See bigspline, bigssa, bigssg, bigssp, and bigtps for more details.

Value

call	Called model in input formula.
type	Type of smoothing spline that was used for each predictor.
fitted.values	Vector of fitted values (if fitresid=TRUE).
linear.predictors	Vector of linear predictors (only for class "bigssg" with fitresid=TRUE).
residuals	Vector of residuals (if fitresid=TRUE). For class "bigssg" these are deviance residuals.
sigma	Estimated error standard deviation.
deviance	Model deviance (only for class "bigssg").
dispersion	Estimated dispersion parameter (only for class "bigssg").
n	Total sample size.
df	Effective degrees of freedom of the model.
info	Model fit information: vector containing the GCV, multiple R-squared, AIC, and BIC of fit model.
converged	Convergence status: converged=TRUE if the iterative theta update converged, converged=FALSE if the iterative theta update failed to converge, and converged=NA if option skip.iter=TRUE was used.
iter	Number of iterative updates (iter=NA if option skip.iter=TRUE was used).
rparm	Rounding parameters used for model fitting.
lambda	Global smoothing parameter used for model fitting.
gammas	Vector of additional smoothing parameters (only for class "bigssa").
thetas	Vector of additional smoothing parameters (only for class "bigssp").
pi	Vector of cosine diagnostics.
family	Distribution family (only for class "bigssg").
gcvtype	Smoothing parameter selection criterion (only for class "bigssg").

Note

For "bigspline" and "bigtps" objects, the outputs call, converged, and iter are NA.

Author(s)

Nathaniel E. Helwig <[e-mail]>

Examples

##########   EXAMPLE 1   ##########

# define relatively smooth function
set.seed(773)
myfun <- function(x){ sin(2*pi*x) }
x <- runif(10^4)
y <- myfun(x) + rnorm(10^4)

# cubic spline
cubmod <- bigspline(x,y)
summary(cubmod)


##########   EXAMPLE 2   ##########

# function with two continuous predictors
set.seed(773)
myfun <- function(x1v,x2v){
  sin(2*pi*x1v) + log(x2v+.1) + cos(pi*(x1v-x2v))
}
x1v <- runif(10^4)
x2v <- runif(10^4)
y <- myfun(x1v,x2v) + rnorm(10^4)

# cubic splines with 100 randomly selected knots (efficient parameterization)
cubmod <- bigssa(y~x1v*x2v,type=list(x1v="cub",x2v="cub"),nknots=100)
summary(cubmod)


##########   EXAMPLE 3   ##########

# function with two continuous predictors
set.seed(1)
myfun <- function(x1v,x2v){
  sin(2*pi*x1v) + log(x2v+.1) + cos(pi*(x1v-x2v))
}
ndpts <- 1000
x1v <- runif(ndpts)
x2v <- runif(ndpts)

# poisson response
set.seed(773)
lp <- myfun(x1v,x2v)
mu <- exp(lp)
y <- rpois(n=ndpts,lambda=mu)

# generalized smoothing spline anova
genmod <- bigssg(y~x1v*x2v,family="poisson",type=list(x1v="cub",x2v="cub"),nknots=50)
summary(genmod)


##########   EXAMPLE 4   ##########

# function with two continuous predictors
set.seed(773)
myfun <- function(x1v,x2v){
  sin(2*pi*x1v) + log(x2v+.1) + cos(pi*(x1v-x2v))
}
x1v <- runif(10^4)
x2v <- runif(10^4)
y <- myfun(x1v,x2v) + rnorm(10^4)

# cubic splines with 100 randomly selected knots (classic parameterization)
cubmod <- bigssp(y~x1v*x2v,type=list(x1v="cub",x2v="cub"),nknots=100)
summary(cubmod)


##########   EXAMPLE 5   ##########

# define relatively smooth function
set.seed(773)
myfun <- function(x){ sin(2*pi*x) }
x <- runif(10^4)
y <- myfun(x) + rnorm(10^4)

# thin-plate with default (30 knots)
tpsmod <- bigtps(x,y)
summary(tpsmod)

bigsplines documentation built on May 2, 2019, 9:27 a.m.