predict.ahazpen: Prediction methods for ahazpen

Description

Compute regression coefficient estimates, linear predictor, cumulative hazard function, or integrated martingale residuals for a fitted penalized semiparametric additive hazards model.

Usage

# S3 method for ahazpen
predict(object, newX, type=c("coef","lp","residuals","cumhaz"),
        lambda=NULL, …)
# S3 method for ahazpen
coef(object, …)

Arguments

object

The result of an ahazpen fit.

newX

New matrix of covariates at which to do predictions. Required unless type="coef".

lambda

Value of lambda for at which predictions are to be made. This argument is required for type="residuals" and type="cumhaz". Since predictions rely on interpolations between lambda values, it is recommended not to use a lambda-value smaller than the minimum of object$lambda.

type

The type of prediction. Options are the regression coefficients ("coef"), the linear predictors ("lp"), the (integrated) martingale residuals ("residuals"), or the cumulative hazard ("cumhaz")

…

For future methods.

Value

For type="coef" and type="lp", a matrix of regression coefficients, respectively linear predictions for each value of the penalty parameter.

For type="residuals", a matrix of (integrated) martingale residuals associated with the nonzero penalized regression coefficients for a regularization parameter equal to $lambda$.

For type="cumhaz", an object with S3 class "cumahaz" based on the regression coefficients estimated for a regularization parameter equal to $lambda$, the object containing:

time

Jump times for the cumulative hazard estimate.

cumhaz

The cumulative hazard estimate.

event

Status at jump times (1 corresponds to death, 0 corresponds to entry/exit).

Details

See the details in predict.ahaz for information on the different types of predictions.

Examples

Run this code

# NOT RUN {
data(sorlie)

set.seed(10101)

# Break ties
time <- sorlie$time+runif(nrow(sorlie))*1e-2

# Survival data + covariates
surv <- Surv(time,sorlie$status)
X <- as.matrix(sorlie[,3:ncol(sorlie)])

# Fit additive hazards regression model w/lasso penalty
fit <- ahazpen(surv, X, dfmax=100)

# Coefficients
beta <- predict(fit,X,lambda=0.08,type="coef")
barplot(as.numeric(beta))

# Linear predictions
linpred <- predict(fit,X,lambda=0.1,type="lp")
riskgrp <- factor(linpred < median(linpred))
plot(survfit(surv~riskgrp))

# Residuals
resid <- predict(fit, X, lambda=0.1, type = "residuals")
par(mfrow = c(1,2))
hist(resid[,1],main=colnames(resid)[1])
hist(resid[,2],main=colnames(resid)[2])

# Cumulative hazard
cumhaz <- predict(fit,X,lambda=0.1,type="cumhaz")
plot(cumhaz)


# }

Run the code above in your browser using DataCamp Workspace