plot_dfbetas: Function to plot the DFBETAs of a fitted coxph object.

In benjamin-w-campbell/coxtools: Tools for Cox Proportional Hazards Models

Description

This function assists the user in assessing model performance according to DFBETAs The two plots produced are designed to help the analyst understand which observations are poorly predicted and how model performance fairs with respect to timing of events.

Usage

plot_dfbetas(coxph_fit = NULL, var_names = NULL)

Arguments

coxph_fit	The output from a fitted "coxph" call.
var_names	A custom vector of variable names that correspond to the variables in the model. Should be ordered as the covariates enter the model on the RHS of the formula.

Value

A ggplot2 object presenting the DFBETAs for the model.

Examples

library(survival)
data("heart")
# Coerce from factor
heart$transplant <- as.numeric(heart$transplant)
# Rescale age
heart$age <- heart$age+48
fit <- coxph(Surv(start, stop, event) ~
              age + transplant +surgery,
             data = heart,
             x = TRUE)

plot_dfbetas(coxph_fit = fit, var_names = c("Age", "Transplant", "Surgery"))

benjamin-w-campbell/coxtools documentation built on May 31, 2019, 10:46 p.m.