wqm_cdfest: Compute Nonparametric CDF Estimates

Description Usage Arguments Details Value

Description

Computes nonparametric estimates of the cumulative distribution function using maximum likelihood. If possible, estimation this is done with a generalized version of the Kaplan-Meier estimate, and otherwise, by using Turnbull's e-m algorithm

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
WQMCDFEST(
  y,
  ny,
  codes,
  weight,
  ty,
  nty,
  tcodes,
  n,
  nstart,
  dscrat,
  scrat,
  iscrat,
  kprint,
  maxit,
  tol,
  maxmsd,
  p,
  q,
  prob,
  sd,
  m,
  pchmax,
  lsd,
  ier
)

Arguments

y

Numeric matrix [n x ny] containing the lower and upper limits for group censored observations.

ny

Number of columns in y (either 1 or 2)

codes

Integer vector of censor codes (see Details)

weight

Vector of observation weights or multiplicities

ty

Numeric matrix containing lower and upper truncation limits

nty

Number of columns in ty

tcodes

Integer vector of truncation codes (see details)

n

Number of rows in y

nstart

If nstart = 0 set automatic start values, otherwise send down nstart values in prob for restart

dscrat

Numeric scratch vector (length = 3 * n + 2)

scrat

Numeric scratch array (length = max((7 * n), maxmsd * (maxmsd - 1) / 2))

iscrat

Integer scratch array (length = 6 * n + 4)

maxit

Maximum number of iterations for s-c algorithm

tol

Desired estimation accuracy (0 < tol < 0.1)

maxmsd

Maximum m for which the full information matrix estimates of the standard errors can be computed

iprint

Print level for debug dump

Details

iprint = 0 for no debug output, if > 0 dump setup and every iprint iteration, iprint = 1 gives the maximum amount of output, iprint = 10 is a good choice to debug

Value

A list of length 6

0 no error 1 n <= 0 on input 2 ny not equal to 1 or 2 3 nty not between 0 and 2 4 tol outside range (0, 0.1) 6 censor code out of range 0 to 4 7 y(i,1) != y(i,2) in a type 1, 2, or 3 observation 8 y(i,1) < y(i,2) in type 4 obs 9 ny=1 but type 4 obs found 10 tcode outside range 1 to 4 11 tcode = 1, 2, or 3 but tyl != tyu 12 backwards truncation interval 12 ty(i,1) < ty(i,2) in type 4 obs 13 only 1 col of truncation values, code 4 found 14 not enough data to estimate distribution (e.g., all right censored observations) 15 nstart>0 does not agree with computed m 16 observation not within the truncation interval *21 product-limit estimate could not be computed directly and maxmsd was too small to allow the full information matrix to be computed an approximation was computed under the assumption that the individual hazard estimates are uncorrelated *22 information matrix not positive definite *23 only one non-zero s probability

* signifies warning message only


Auburngrads/SMRD documentation built on Sept. 14, 2020, 2:21 a.m.