icendata {npsurv}

Class of Interval-censored Data

Description

Class icendata can be used to store general interval-censored data, which may possibly contain exact observations.There are several functions associated with the class.

Usage

icendata(x, w=1)
is.icendata(x)

Arguments

x

vector or matrix.

w

weights or multiplicities of the observations.

Details

Function icendata creates an object of class 'icendata', which can be used to save both interval-censored and exact observations.

Function is.icendata simply checks if an object is of class 'icendata'.

If x is a vector, it contains only exact observations, with weights given in w.

If x is a two-column matrix, it contains interval-censored observations and stores their left and right endpoints in the first and second column, respectively. If the left and right endpoints are equal, then the observation is exact. Weights are provided by w.

If x is a three-column matrix, it contains interval-censored observations and stores their left and right endpoints in the first and second column, respectively. The weight of each observation is the third-column value multiplied by the corresponding weight value in w.

It is useful to turn interval-censored (and exact) observations into the format imposed by icendata so that they can be processed in a standardized format by other functions. Also, exact and interval-censored observations are stored separately in this format and can hence be dealt with more easily. Most functions in the package npsurv first ensure that the data has this format before processing.

Observations of zero weights are removed. Identical observations are aggregated.

An interval-valued observation is either (L_i, R_i] if L_i < R_i, or [L_i, R_i] if L_i = R_i.

Value

t

numeric vector, storing exact observations.

wt

numeric vector, storing the weights of exact observations.

o

two-column numeric matrix, storing interval-censored observations.

wo

numeric vector, storing the weights of interval-censored observations.

i1

logical vector, indicating whether exact observations are less than upper.

upper

the largest finite value of t and o.

u

numeric vector, containing 0 and all unique finite values in t and o.

Author(s)

Yong Wang <[e-mail]>

References

Wang, Y. (2008). Dimension-reduced nonparametric maximum likelihood computation for interval-censored data. Computational Statistics & Data Analysis, 52, 2388-2402.

Wang, Y. and Fani, S. (2017). Nonparametric maximum likelihood computation of a U-shaped hazard function. Statistics and Computing, (in print).

See Also

npsurv, Uhaz.

Examples

data(ap)
(x = icendata(ap))
is.icendata(x)

data(gastric)
icendata(gastric)

data(leukemia)
i = leukemia[,"group"] == "6-MP"
icendata(leukemia[i,1:2])