BSBInit {BGPhazard}R Documentation

Initial setup for BSBHaz model


BSBInit creates the necessary data structure for use in BSBHaz.


  df = NULL,
  t1 = NULL,
  t2 = NULL,
  alpha = 0.001,
  beta = 0.001,
  c = 1000,
  part_len = 1,
  seed = 42



A data frame with columns 't1', 't2', 'delta1', 'delta2'. Any other columns not named 'id' are taken to be predictors. These predictors must be numeric, i.e., categorical predictors must be one-hot encoded.

t1, t2

Objects of class 'Surv' as created by Surv.

alpha, beta, c

Doubles. Parameters for Markov gamma hazard priors.


A double that gives the length of time partition intervals.


Random seed for variable initialization.


This function reads and formats censored bivariate survival data in the following way. If df is provided, failure times and censoring indicadors are assumed to be columns named 't1', 't2', 'delta1', and 'delta2'. Other columns not named 'id' (ignoring case) are taken to be predictors. If df has no columns 'delta1' or 'delta2', observations are taken as exact.

If df is not provided, then t1 and t2 are expected to be objects of class 'Surv' created by Surv and the model does not use predictors. Only right-censored observations are supported. Only df or t1 and t2 must be supplied. df argument comes first for use in pipes.


An object of class 'BSBinit'


t1 <- survival::Surv(c(1, 2, 3))
t2 <- survival::Surv(c(1, 2, 3))

init <- BSBInit(t1 = t1, t2 = t2, seed = 0)

[Package BGPhazard version 2.1.1 Index]