Discrete Response Regression for High-Dimensional Data: Discrete Response Generalized Monotone Incremental Forward Stagewise Regression

Description

This package provides a function that fits a Poisson or negative binomial model when the number of parameters exceeds the sample size, using the the generalized monotone incremental forward stagewise method.

Details

The DESCRIPTION file:

Index of help topics:

coef.countgmifs         Extract Model Coefficients.
countgmifs              Discrete Response Generalized Monotone
                        Incremental Forward Stagewise Regression.
countgmifs-package      Discrete Response Regression for
                        High-Dimensional Data: Discrete Response
                        Generalized Monotone Incremental Forward
                        Stagewise Regression
plot.countgmifs         Plot Solution Path for a Count GMIFS Fitted
                        Model.
predict.countgmifs      Predict Outcome for Count GMIFS Fitted Model.
print.countgmifs        Print the Contents of a Count GMIFS Fitted
                        Object.
summary.countgmifs      Summarize a Count GMIFS Object.

This package contains functions for fitting a penalized discrete response model (either negative binomial or Poisson) and extracting estimated coefficients, predictions, and plots. The model and methods can be used when the response to be predicted is discrete, and is particularly relevant when there are more covariates than observations.

Author(s)

NA Kellie J. Archer <archer.43@osu.edu>

Maintainer: NA Kellie J. Archer <archer.43@osu.edu>

References

Makowski M., Archer K.J. (2015) Generalized monotone incremental forward stagewise method for modeling count data: application predicting micronuclei frequency. Cancer Informatics, 14(Suppl 2), 97–105.

Extract Model Coefficients.

Description

A generic function which extracts the model coefficients from a fitted model object fit using countgmifs

Usage

## S3 method for class 'countgmifs'
coef(object, model.select = "BIC", ...)

Arguments

See Also

See Also countgmifs, predict.countgmifs, summary.countgmifs, plot.countgmifs

Discrete Response Generalized Monotone Incremental Forward Stagewise Regression.

Description

This function can fit a Poisson or negative binomial model when the number of parameters exceeds the sample size, using the the generalized monotone incremental forward stagewise method.

Usage

countgmifs(formula, data, x = NULL, offset, subset, epsilon = 0.001,
  tol = 1e-05, scale = TRUE, verbose = FALSE, family = "nb", ...)

Arguments

See Also

See Also coef.countgmifs, summary.countgmifs, predict.countgmifs, plot.countgmifs

Examples

set.seed(26)
n <- 50 # Sample size
p <- 500 # Number of covariates
intercept<- .5
#True parameter values for the 500 covariates
beta<- c(log(1.5), log(1.5), -log(1.5), -log(1.5), -log(1.5), rep(0,495)) 
alpha<- 0.5 # Intercept
x<- matrix(rnorm(n*p,0,1), nrow=n, ncol=p, byrow=TRUE) #Covariate values
colnames(x)<- paste("Var",1:p, sep="")  
mu<- exp(intercept + crossprod(t(x),beta))
y<- rnbinom(n=n, size=1/alpha ,mu=mu) # Discrete response
data<- data.frame(y,x)
nb<-countgmifs(y ~ 1 , data=data, offset=NULL, x=x, epsilon=0.01, tol=0.001, 
       scale=TRUE, verbose=FALSE)
coef.AIC<-coef(nb, model.select="AIC")
coef.AIC[coef.AIC!=0]
predict(nb, model.select="AIC")
plot(predict(nb, model.select="AIC"), y)
plot(nb)

Functions Called by countgmifs Functions, Not by the User

Description

These functions are called my other countgmifs functions and are not intended to be directly called by the user.

Details

The nb.theta and poisson.theta functions for the negative binomial and poisson regression models, respectively, are the log-likelihood functions and are used to estimate the intercept non-penalized subset parameters (if included) at each step of the algorithm. The hilbe function Hilbes estimate of the alpha parameter for the negative binomial regression model.

Author(s)

Kellie J. Archer, archer.43@osu.edu

See Also

See Also as countgmifs

Plot Solution Path for a Count GMIFS Fitted Model.

Description

This function plots either the coefficient path, the AIC, or the log-likelihood for a fitted countgmifs object.

Usage

## S3 method for class 'countgmifs'
plot(x, type = "trace", xlab = NULL, ylab = NULL,
  main = NULL, ...)

Arguments

See Also

See Also countgmifs, coef.countgmifs, summary.countgmifs, predict.countgmifs

Predict Outcome for Count GMIFS Fitted Model.

Description

This function returns a numeric vector that is the predicted response from the countgmifs fitted object.

Usage

## S3 method for class 'countgmifs'
predict(object, neww = NULL, newdata, newx = NULL,
  model.select = "BIC", newoffset=NULL, ...)

Arguments

See Also

See Also countgmifs, coef.countgmifs, summary.countgmifs, plot.countgmifs

Print the Contents of a Count GMIFS Fitted Object.

Description

This function prints the names of the list objects from an countgmifs fitted model

Usage

## S3 method for class 'countgmifs'
print(x, ...)

Arguments

See Also

See Also countgmifs, coef.countgmifs, summary.countgmifs, plot.countgmifs

Summarize a Count GMIFS Object.

Description

Prints the following items extracted from the fitted countgmifs object: the family used and model parameter estimates. For models that include x, the parameter estimates, AIC, BIC, and log-likelihood are printed for indicated model.select step or if model.select is not supplied the step at which the minimum BIC was observed.

Usage

## S3 method for class 'countgmifs'
summary(object, model.select = "BIC", ...)

Arguments

See Also

See Also countgmifs, coef.countgmifs, predict.countgmifs, plot.countgmifs