ic_sample function

Draw samples from an icenReg model

Samples response values from an icenReg fit conditional on covariates, but not censoring intervals. To draw response samples conditional on covariates and restrained to intervals, see imputeCens.

ic_sample(fit, newdata = NULL, sampleType = "fullSample", samples = 5)

Arguments

• fit: icenReg model fit
• newdata: data.frame containing covariates. If blank, will use data from model
• sampleType: type of samples See details for options
• samples: Number of samples

Details

Returns a matrix of samples. Each row of the matrix corresponds with a subject with the covariates of the corresponding row of newdata. For each column of the matrix, the same sampled parameters are used to sample response variables.

If newdata is left blank, will provide estimates for original data set.

There are several options for how to sample. To get random samples without accounting for error in the estimated parameters imputeType ='fixedParSample' takes a random sample of the response variable, conditional on the response interval, covariates and estimated parameters at the MLE. Alternatively, imputeType = 'fullSample' first takes a random sample of the coefficients, (assuming asymptotic normality for the ic_par) and then takes a random sample of the response variable, conditional on the response interval, covariates, and the random sample of the coefficients.

Examples

simdata <- simIC_weib(n = 500)

fit <- ic_par(cbind(l, u) ~ x1 + x2,
              data = simdata)

newdata = data.frame(x1 = c(0, 1), x2 = c(1,1))

sampleResponses <- ic_sample(fit, newdata = newdata, samples = 100)

Author(s)

Clifford Anderson-Bergman