g08 Chapter Contents
g08 Chapter Introduction
NAG C Library Manual

# NAG Library Function Documentnag_rank_regsn_censored (g08rbc)

## 1  Purpose

nag_rank_regsn_censored (g08rbc) calculates the parameter estimates, score statistics and their variance-covariance matrices for the linear model using a likelihood based on the ranks of the observations when some of the observations may be right-censored.

## 2  Specification

 #include #include
 void nag_rank_regsn_censored (Nag_OrderType order, Integer ns, const Integer nv[], const double y[], Integer p, const double x[], Integer pdx, const Integer icen[], double gamma, Integer nmax, double tol, double prvr[], Integer pdprvr, Integer irank[], double zin[], double eta[], double vapvec[], double parest[], NagError *fail)

## 3  Description

Analysis of data can be made by replacing observations by their ranks. The analysis produces inference for the regression model where the location parameters of the observations, ${\theta }_{\mathit{i}}$, for $\mathit{i}=1,2,\dots ,n$, are related by $\theta =X\beta$. Here $X$ is an $n$ by $p$ matrix of explanatory variables and $\beta$ is a vector of $p$ unknown regression parameters. The observations are replaced by their ranks and an approximation, based on Taylor's series expansion, made to the rank marginal likelihood. For details of the approximation see Pettitt (1982).
An observation is said to be right-censored if we can only observe ${Y}_{j}^{*}$ with ${Y}_{j}^{*}\le {Y}_{j}$. We rank censored and uncensored observations as follows. Suppose we can observe ${Y}_{\mathit{j}}$, for $\mathit{j}=1,2,\dots ,n$, directly but ${Y}_{j}^{*}$, for $\mathit{j}=n+1,\dots ,q$ and $n\le q$, are censored on the right. We define the rank ${r}_{j}$ of ${Y}_{j}$, for $j=1,2,\dots ,n$, in the usual way; ${r}_{j}$ equals $i$ if and only if ${Y}_{j}$ is the $i$th smallest amongst the ${Y}_{1},{Y}_{2},\dots ,{Y}_{n}$. The right-censored ${Y}_{j}^{*}$, for $j=n+1,n+2,\dots ,q$, has rank ${r}_{j}$ if and only if ${Y}_{j}^{*}$ lies in the interval $\left[{Y}_{\left({r}_{j}\right)},{Y}_{\left({r}_{j}+1\right)}\right]$, with ${Y}_{0}=-\infty$, ${Y}_{\left(n+1\right)}=+\infty$ and ${Y}_{\left(1\right)}<\cdots <{Y}_{\left(n\right)}$ the ordered ${Y}_{\mathit{j}}$, for $\mathit{j}=1,2,\dots ,n$.
The distribution of the $Y$ is assumed to be of the following form. Let ${F}_{L}\left(y\right)={e}^{y}/\left(1+{e}^{y}\right)$, the logistic distribution function, and consider the distribution function ${F}_{\gamma }\left(y\right)$ defined by $1-{F}_{\gamma }={\left[1-{F}_{L}\left(y\right)\right]}^{1/\gamma }$. This distribution function can be thought of as either the distribution function of the minimum, ${X}_{1,\gamma }$, of a random sample of size ${\gamma }^{-1}$ from the logistic distribution, or as the ${F}_{\gamma }\left(y-\mathrm{log}\gamma \right)$ being the distribution function of a random variable having the $F$-distribution with $2$ and $2{\gamma }^{-1}$ degrees of freedom. This family of generalized logistic distribution functions $\left[{F}_{\gamma }\left(.\right)\text{;}0\le \gamma <\infty \right]$ naturally links the symmetric logistic distribution $\left(\gamma =1\right)$ with the skew extreme value distribution ($\mathrm{lim}\gamma \to 0$) and with the limiting negative exponential distribution ($\mathrm{lim}\gamma \to \infty$). For this family explicit results are available for right-censored data. See Pettitt (1983) for details.
Let ${l}_{R}$ denote the logarithm of the rank marginal likelihood of the observations and define the $q×1$ vector $a$ by $a={l}_{R}^{\prime }\left(\theta =0\right)$, and let the $q$ by $q$ diagonal matrix $B$ and $q$ by $q$ symmetric matrix $A$ be given by $B-A=-{l}_{R}^{\prime \prime }\left(\theta =0\right)$. Then various statistics can be found from the analysis.
 (a) The score statistic ${X}^{\mathrm{T}}a$. This statistic is used to test the hypothesis ${H}_{0}:\beta =0$ (see (e)). (b) The estimated variance-covariance matrix of the score statistic in (a). (c) The estimate ${\stackrel{^}{\beta }}_{R}=M{X}^{\mathrm{T}}a$. (d) The estimated variance-covariance matrix $M={\left({X}^{\mathrm{T}}\left(B-A\right)X\right)}^{-1}$ of the estimate ${\stackrel{^}{\beta }}_{R}$. (e) The ${\chi }^{2}$ statistic $Q={\stackrel{^}{\beta }}_{R}{M}^{-1}\text{​ ​}{\stackrel{^}{\beta }}_{r}={a}^{\mathrm{T}}X{\left({X}^{\mathrm{T}}\left(B-A\right)X\right)}^{-1}{X}^{\mathrm{T}}a$, used to test ${H}_{0}:\beta =0$. Under ${H}_{0}$, $Q$ has an approximate ${\chi }^{2}$-distribution with $p$ degrees of freedom. (f) The standard errors ${M}_{ii}^{1/2}$ of the estimates given in (c). (g) Approximate $z$-statistics, i.e., ${Z}_{i}={\stackrel{^}{\beta }}_{{R}_{i}}/se\left({\stackrel{^}{\beta }}_{{R}_{i}}\right)$ for testing ${H}_{0}:{\beta }_{i}=0$. For $i=1,2,\dots ,n$, ${Z}_{i}$ has an approximate $N\left(0,1\right)$ distribution.
In many situations, more than one sample of observations will be available. In this case we assume the model,
 $hk Yk = XkT β+ek , k=1,2,…,ns ,$
where ns is the number of samples. In an obvious manner, ${Y}_{k}$ and ${X}_{k}$ are the vector of observations and the design matrix for the $k$th sample respectively. Note that the arbitrary transformation ${h}_{k}$ can be assumed different for each sample since observations are ranked within the sample.
The earlier analysis can be extended to give a combined estimate of $\beta$ as $\stackrel{^}{\beta }=Dd$, where
 $D-1=∑k=1nsXTBk-AkXk$
and
 $d=∑k= 1ns XkT ak ,$
with ${a}_{k}$, ${B}_{k}$ and ${A}_{k}$ defined as $a$, $B$ and $A$ above but for the $k$th sample.
The remaining statistics are calculated as for the one sample case.

## 4  References

Kalbfleisch J D and Prentice R L (1980) The Statistical Analysis of Failure Time Data Wiley
Pettitt A N (1982) Inference for the linear model using a likelihood based on ranks J. Roy. Statist. Soc. Ser. B 44 234–243
Pettitt A N (1983) Approximate methods using ranks for regression with censored data Biometrika 70 121–132

## 5  Arguments

1:     orderNag_OrderTypeInput
On entry: the order argument specifies the two-dimensional storage scheme being used, i.e., row-major ordering or column-major ordering. C language defined storage is specified by ${\mathbf{order}}=\mathrm{Nag_RowMajor}$. See Section 3.2.1.3 in the Essential Introduction for a more detailed explanation of the use of this argument.
Constraint: ${\mathbf{order}}=\mathrm{Nag_RowMajor}$ or Nag_ColMajor.
2:     nsIntegerInput
On entry: the number of samples.
Constraint: ${\mathbf{ns}}\ge 1$.
3:     nv[ns]const IntegerInput
On entry: the number of observations in the $\mathit{i}$th sample, for $\mathit{i}=1,2,\dots ,{\mathbf{ns}}$.
Constraint: ${\mathbf{nv}}\left[\mathit{i}-1\right]\ge 1$, for $\mathit{i}=1,2,\dots ,{\mathbf{ns}}$.
4:     y[$\mathit{dim}$]const doubleInput
Note: the dimension, dim, of the array y must be at least $\left(\sum _{\mathit{i}=1}^{{\mathbf{ns}}}{\mathbf{nv}}\left[\mathit{i}-1\right]\right)$.
On entry: the observations in each sample. Specifically, ${\mathbf{y}}\left[\sum _{k=1}^{i-1}{\mathbf{nv}}\left[k-1\right]+j-1\right]$ must contain the $j$th observation in the $i$th sample.
5:     pIntegerInput
On entry: the number of parameters to be fitted.
Constraint: ${\mathbf{p}}\ge 1$.
6:     x[$\mathit{dim}$]const doubleInput
Note: the dimension, dim, of the array x must be at least
• $\mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(1,{\mathbf{pdx}}×{\mathbf{p}}\right)$ when ${\mathbf{order}}=\mathrm{Nag_ColMajor}$;
• $\mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(1,\left(\sum _{\mathit{i}=1}^{{\mathbf{ns}}}{\mathbf{nv}}\left[\mathit{i}-1\right]\right)×{\mathbf{pdx}}\right)$ when ${\mathbf{order}}=\mathrm{Nag_RowMajor}$.
Where ${\mathbf{X}}\left(i,j\right)$ appears in this document, it refers to the array element
• ${\mathbf{x}}\left[\left(j-1\right)×{\mathbf{pdx}}+i-1\right]$ when ${\mathbf{order}}=\mathrm{Nag_ColMajor}$;
• ${\mathbf{x}}\left[\left(i-1\right)×{\mathbf{pdx}}+j-1\right]$ when ${\mathbf{order}}=\mathrm{Nag_RowMajor}$.
On entry: the design matrices for each sample. Specifically, ${\mathbf{X}}\left(\sum _{k=1}^{i-1}{\mathbf{nv}}\left[k-1\right]+j,l\right)$ must contain the value of the $l$th explanatory variable for the $j$th observations in the $i$th sample.
Constraint: ${\mathbf{x}}$ must not contain a column with all elements equal.
7:     pdxIntegerInput
On entry: the stride separating row or column elements (depending on the value of order) in the array x.
Constraints:
• if ${\mathbf{order}}=\mathrm{Nag_ColMajor}$, ${\mathbf{pdx}}\ge \left(\sum _{\mathit{i}=1}^{{\mathbf{ns}}}{\mathbf{nv}}\left[\mathit{i}-1\right]\right)$;
• if ${\mathbf{order}}=\mathrm{Nag_RowMajor}$, ${\mathbf{pdx}}\ge {\mathbf{p}}$.
8:     icen[$\mathit{dim}$]const IntegerInput
Note: the dimension, dim, of the array icen must be at least $\left(\sum _{\mathit{i}=1}^{{\mathbf{ns}}}{\mathbf{nv}}\left[\mathit{i}-1\right]\right)$.
On entry: defines the censoring variable for the observations in y.
${\mathbf{icen}}\left[i-1\right]=0$
If ${\mathbf{y}}\left[i-1\right]$ is uncensored.
${\mathbf{icen}}\left[i-1\right]=1$
If ${\mathbf{y}}\left[i-1\right]$ is censored.
Constraint: ${\mathbf{icen}}\left[\mathit{i}-1\right]=0$ or $1$, for $\mathit{i}=1,2,\dots ,\left(\sum _{\mathit{i}=1}^{{\mathbf{ns}}}{\mathbf{nv}}\left[\mathit{i}-1\right]\right)$.
On entry: the value of the parameter defining the generalized logistic distribution. For ${\mathbf{gamma}}\le 0.0001$, the limiting extreme value distribution is assumed.
Constraint: ${\mathbf{gamma}}\ge 0.0$.
10:   nmaxIntegerInput
On entry: the value of the largest sample size.
Constraint: ${\mathbf{nmax}}=\underset{1\le i\le {\mathbf{ns}}}{\mathrm{max}}\phantom{\rule{0.25em}{0ex}}\left({\mathbf{nv}}\left[i-1\right]\right)$ and ${\mathbf{nmax}}>{\mathbf{p}}$.
11:   toldoubleInput
On entry: the tolerance for judging whether two observations are tied. Thus, observations ${Y}_{i}$ and ${Y}_{j}$ are adjudged to be tied if $\left|{Y}_{i}-{Y}_{j}\right|<{\mathbf{tol}}$.
Constraint: ${\mathbf{tol}}>0.0$.
12:   prvr[$\mathit{dim}$]doubleOutput
Note: the dimension, dim, of the array prvr must be at least
• $\mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(1,{\mathbf{pdprvr}}×{\mathbf{p}}\right)$ when ${\mathbf{order}}=\mathrm{Nag_ColMajor}$;
• $\mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(1,{\mathbf{p}}+1×{\mathbf{pdprvr}}\right)$ when ${\mathbf{order}}=\mathrm{Nag_RowMajor}$.
Where ${\mathbf{PRVR}}\left(i,j\right)$ appears in this document, it refers to the array element
• ${\mathbf{prvr}}\left[\left(j-1\right)×{\mathbf{pdprvr}}+i-1\right]$ when ${\mathbf{order}}=\mathrm{Nag_ColMajor}$;
• ${\mathbf{prvr}}\left[\left(i-1\right)×{\mathbf{pdprvr}}+j-1\right]$ when ${\mathbf{order}}=\mathrm{Nag_RowMajor}$.
On exit: the variance-covariance matrices of the score statistics and the parameter estimates, the former being stored in the upper triangle and the latter in the lower triangle. Thus for $1\le i\le j\le {\mathbf{p}}$, ${\mathbf{PRVR}}\left(i,j\right)$ contains an estimate of the covariance between the $i$th and $j$th score statistics. For $1\le j\le i\le {\mathbf{p}}-1$, ${\mathbf{PRVR}}\left(i+1,j\right)$ contains an estimate of the covariance between the $i$th and $j$th parameter estimates.
13:   pdprvrIntegerInput
On entry: the stride separating row or column elements (depending on the value of order) in the array prvr.
Constraints:
• if ${\mathbf{order}}=\mathrm{Nag_ColMajor}$, ${\mathbf{pdprvr}}\ge {\mathbf{p}}+1$;
• if ${\mathbf{order}}=\mathrm{Nag_RowMajor}$, ${\mathbf{pdprvr}}\ge {\mathbf{p}}$.
14:   irank[nmax]IntegerOutput
On exit: for the one sample case, irank contains the ranks of the observations.
15:   zin[nmax]doubleOutput
On exit: for the one sample case, zin contains the expected values of the function $g\left(.\right)$ of the order statistics.
16:   eta[nmax]doubleOutput
On exit: for the one sample case, eta contains the expected values of the function $g\prime \left(.\right)$ of the order statistics.
17:   vapvec[${\mathbf{nmax}}×\left({\mathbf{nmax}}+1\right)/2$]doubleOutput
On exit: for the one sample case, vapvec contains the upper triangle of the variance-covariance matrix of the function $g\left(.\right)$ of the order statistics stored column-wise.
18:   parest[$4×{\mathbf{p}}+1$]doubleOutput
On exit: the statistics calculated by the function.
The first p components of parest contain the score statistics.
The next p elements contain the parameter estimates.
${\mathbf{parest}}\left[2×{\mathbf{p}}\right]$ contains the value of the ${\chi }^{2}$ statistic.
The next p elements of parest contain the standard errors of the parameter estimates.
Finally, the remaining p elements of parest contain the $z$-statistics.
19:   failNagError *Input/Output
The NAG error argument (see Section 3.6 in the Essential Introduction).

## 6  Error Indicators and Warnings

NE_ALLOC_FAIL
Dynamic memory allocation failed.
On entry, argument $〈\mathit{\text{value}}〉$ had an illegal value.
NE_INT
On entry, ${\mathbf{ns}}=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{ns}}\ge 1$.
On entry, ${\mathbf{p}}=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{p}}\ge 1$.
On entry, ${\mathbf{pdprvr}}=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{pdprvr}}>0$.
On entry, ${\mathbf{pdx}}=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{pdx}}>0$.
NE_INT_2
On entry, ${\mathbf{nmax}}=〈\mathit{\text{value}}〉$ and ${\mathbf{p}}=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{nmax}}>{\mathbf{p}}$.
On entry, ${\mathbf{pdprvr}}=〈\mathit{\text{value}}〉$ and ${\mathbf{p}}=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{pdprvr}}\ge {\mathbf{p}}$.
On entry, ${\mathbf{pdprvr}}=〈\mathit{\text{value}}〉$ and ${\mathbf{p}}=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{pdprvr}}\ge {\mathbf{p}}+1$.
On entry, ${\mathbf{pdx}}=〈\mathit{\text{value}}〉$ and ${\mathbf{p}}=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{pdx}}\ge {\mathbf{p}}$.
On entry, ${\mathbf{pdx}}=〈\mathit{\text{value}}〉$ and sum ${\mathbf{nv}}\left[i-1\right]=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{pdx}}\ge \text{}$ the sum of ${\mathbf{nv}}\left[i-1\right]$.
NE_INT_ARRAY
On entry, ${\mathbf{ns}}=〈\mathit{\text{value}}〉$ and ${\mathbf{nv}}\left[\mathit{i}-1\right]=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{nv}}\left[\mathit{i}-1\right]\ge 1$, for $\mathit{i}=1,2,\dots ,{\mathbf{ns}}$.
NE_INT_ARRAY_ELEM_CONS
On entry $M=〈\mathit{\text{value}}〉$.
Constraint: $M$ elements of array ${\mathbf{icen}}=0$ or $1$.
On entry $M=〈\mathit{\text{value}}〉$.
Constraint: $M$ elements of array ${\mathbf{nv}}>0$.
NE_INTERNAL_ERROR
An internal error has occurred in this function. Check the function call and any array sizes. If the call is correct then please contact NAG for assistance.
NE_MAT_ILL_DEFINED
The matrix ${X}^{\mathrm{T}}\left(B-A\right)X$ is either singular or non positive definite.
NE_OBSERVATIONS
All the observations were adjudged to be tied.
NE_REAL
On entry, ${\mathbf{gamma}}=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{gamma}}\ge 0.0$.
On entry, ${\mathbf{tol}}=〈\mathit{\text{value}}〉$.
Constraint: ${\mathbf{tol}}>0.0$.
NE_REAL_ARRAY_ELEM_CONS
On entry, all elements in column $〈\mathit{\text{value}}〉$ of ${\mathbf{x}}$ are equal to $〈\mathit{\text{value}}〉$.
NE_SAMPLE
The largest sample size is $〈\mathit{\text{value}}〉$ which is not equal to nmax, ${\mathbf{nmax}}=〈\mathit{\text{value}}〉$.

## 7  Accuracy

The computations are believed to be stable.

The time taken by nag_rank_regsn_censored (g08rbc) depends on the number of samples, the total number of observations and the number of parameters fitted.
In extreme cases the parameter estimates for certain models can be infinite, although this is unlikely to occur in practice. See Pettitt (1982) for further details.

## 9  Example

This example fits a regression model to a single sample of $40$ observations using just one explanatory variable.

### 9.1  Program Text

Program Text (g08rbce.c)

### 9.2  Program Data

Program Data (g08rbce.d)

### 9.3  Program Results

Program Results (g08rbce.r)