predict.glmpathcr {glmpathcr}

Predicted Class and Fitted Probabilities from glmpathcr Object

Description

Returns either the predicted class or the fitted probabilities from the penalized continuation ratio model from a glmpathcr object.

Usage

## S3 method for class 'glmpathcr'
predict(object, newx = NULL, which = "BIC", type = "class", ...)
## S3 method for class 'glmpathcr'
fitted(object, newx = NULL, which = "BIC", type = "class", ...)

Arguments

object

a glmpathcr object

newx

a data matrix containing the predictor variables, if NULL, the original training data is used for getting model predictions

which

used for model selection, either the default which="BIC" or which="AIC"

type

used for selecting desired output, either the default type="class" or type="probes" for the fitted probabilities

...

additional optional arguments

Value

Either a vector of the predicted class labels or a matrix containing the fitted probabilities is returned.

Author(s)

Kellie J. Archer, [e-mail]

References

Archer K.J., Williams A.A.A. (2012) L1 penalized continuation ratio models for ordinal response prediction using high-dimensional datasets. Statistics in Medicine, 31(14), 1464-74.

See Also

See also as glmpathcr, coef.glmpathcr.

Examples

data(diabetes)
x <- diabetes[, 2:dim(diabetes)[2]]
y <- diabetes$y
fit <- glmpathcr(x, y)
pred <- predict(fit)
table(pred, y)
pred <- predict(fit, which="AIC", type="probs")