Fits the non-parametric maximum likelihood estimator (NPMLE) for univariate interval censored data. This is a generalization of the Kaplan-Meier curves that allows for interval censoring. Also referred to as the Turnbull estimator.
ic_np(
formula = NULL,
data,
maxIter = 1000,
tol = 10^-10,
B = c(0, 1),
weights = NULL
)Formula for stratification. If only one group, can be left blank and data must be entered as n x 2 matrix.
A data.frame or an n x 2 matrix. See details.
Maximum iterations
Numeric tolerance
Should intervals be open or closed? See details.
Weights (optional)
Clifford Anderson-Bergman
data must be an n x 2 matrix or data.frame containing two columns of data representing
left and right sides of the censoring interval, denoted L and R. This allows for left censored
(L == 0), right censored (R == inf), uncensored (L == R) along with general interval censored observations.
The argument B determines whether the intervals should be open or closed, i.e.
B = c(0,1) implies that the event occurs within the interval (l,u].
The exception is that if l == u, it is assumed that the event is uncensored, regardless of B.
The NPMLE is fit using an efficient implementation of the EMICM algorithm.
Turnbull, B. (1976) The empricial distribution with arbitrarily grouped and censored data Journal of the Royal Statistical Society B, vol 38 p290-295
Wellner, J. A., and Zhan, Y. (1997) A hybrid algorithm for computation of the maximum likelihood estimator from censored data, Journal of the American Statistical Association, Vol 92, pp945-959
Anderson-Bergman, C. (2016) An efficient implementation of the EMICM algorithm for the interval censored NPMLE Journal of Computational and Graphical Statistics, just accepted
data(miceData)
fit <- ic_np(cbind(l, u) ~ grp, data = miceData)
# Stratifies fits by group
plot(fit)
Run the code above in your browser using DataLab