apricom (version 1.0.0)

kcrossval: Cross-validation-derived Shrinkage After Estimation

Description

Shrink regression coefficients using a Cross-validation-derived shrinkage factor.

Usage

kcrossval(dataset, model, k, nreps, sdm, int = TRUE, int.adj)

Arguments

dataset

a dataset for regression analysis. Data should be in the form of a matrix, with the outcome variable as the final column. Application of the datashape function beforehand is recommended, especially if categorical predictors are present. For regression with an intercept included a column vector of 1s should be included before the dataset (see examples)

model

type of regression model. Either "linear" or "logistic".

k

the number of cross-validation folds. This number must be within the range 1 < k

nreps

the number of times to replicate the cross-validation process.

sdm

a shrinkage design matrix. For examples, see ols.shrink

int

logical. If TRUE the model will include a regression intercept.

int.adj

logical. If TRUE the regression intercept will be re-estimated after shrinkage of the regression coefficients.

Value

kcrossval returns a list containing the following:

raw.coeff

the raw regression model coefficients, pre-shrinkage.

shrunk.coeff

the shrunken regression model coefficients.

lambda

the mean shrinkage factor over nreps cross-validation replicates.

nFolds

the number of cross-validation folds.

nreps

the number of cross-validation replicates.

sdm

the shrinkage design matrix used to apply the shrinkage factor(s) to the regression coefficients.

Details

This function applies k-fold cross-validation to a dataset in order to derive a shrinkage factor and apply it to the regression coefficients. Data is randomly partitioned into k equally sized sets. One set is used as a validation set, while the remaining data is used as a training set. Regression coefficients are estimated in the training set, and then a shrinkage factor is estimated using the validation set. This process is repeated so that each partitioned set is used as the validation set once, resulting in k folds. The mean of k shrinkage factors is then applied to the original regression coeffients, and the regression intercept may be re-estimated. This process is repeated nreps times and the mean of the regression coefficients is returned.

This process can currently be applied to linear or logistic regression models.

Examples

## Example 1: Linear regression using the iris dataset
## 2-fold Cross-validation-derived shrinkage with 50 reps
data(iris)
iris.data <- as.matrix(iris[, 1:4])
iris.data <- cbind(1, iris.data)
sdm1 <- matrix(c(0, 1, 1, 1), nrow = 1)
kcrossval(dataset = iris.data, model = "linear", k = 2,
nreps = 50, sdm = sdm1, int = TRUE, int.adj = TRUE)

## Example 2: logistic regression using a subset of the mtcars data
## 10-fold CV-derived shrinkage (uniform shrinkage and intercept re-estimation)
data(mtcars)
mtc.data <- cbind(1,datashape(mtcars, y = 8, x = c(1, 6, 9)))
head(mtc.data)
set.seed(321)
kcrossval(dataset = mtc.data, model = "logistic",
k = 10, nreps = 10)