NAG Library Routine Document

g08raf  (rank_regsn)

 Contents

    1  Purpose
    7  Accuracy

1
Purpose

g08raf 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.

2
Specification

Fortran Interface
Subroutine g08raf ( ns, nv, nsum, y, ip, x, ldx, idist, nmax, tol, prvr, ldprvr, irank, zin, eta, vapvec, parest, work, lwork, iwa, ifail)
Integer, Intent (In):: ns, nsum, ip, ldx, idist, nmax, ldprvr, lwork
Integer, Intent (Inout):: nv(ns), ifail
Integer, Intent (Out):: irank(nmax), iwa(nmax)
Real (Kind=nag_wp), Intent (In):: y(nsum), x(ldx,ip), tol
Real (Kind=nag_wp), Intent (Inout):: prvr(ldprvr,ip)
Real (Kind=nag_wp), Intent (Out):: zin(nmax), eta(nmax), vapvec(nmax*(nmax+1)/2), parest(4*ip+1), work(lwork)
C Header Interface
#include nagmk26.h
void  g08raf_ ( const Integer *ns, Integer nv[], const Integer *nsum, const double y[], const Integer *ip, const double x[], const Integer *ldx, const Integer *idist, const Integer *nmax, const double *tol, double prvr[], const Integer *ldprvr, Integer irank[], double zin[], double eta[], double vapvec[], double parest[], double work[], const Integer *lwork, Integer iwa[], Integer *ifail)

3
Description

Analysis of data can be made by replacing observations by their ranks. The analysis produces inference for regression arguments arising from the following model.
For random variables Y1,Y2,,Yn we assume that, after an arbitrary monotone increasing differentiable transformation, h., the model
hYi= xiT β+εi (1)
holds, where xi is a known vector of explanatory variables and β is a vector of p unknown regression coefficients. The εi are random variables assumed to be independent and identically distributed with a completely known distribution which can be one of the following: Normal, logistic, extreme value or double-exponential. In Pettitt (1982) an estimate for β is proposed as β^=MXTa with estimated variance-covariance matrix M. The statistics a and M depend on the ranks ri of the observations Yi and the density chosen for εi.
The matrix X is the n by p matrix of explanatory variables. It is assumed that X is of rank p and that a column or a linear combination of columns of X is not equal to the column vector of 1 or a multiple of it. This means that a constant term cannot be included in the model (1). The statistics a and M are found as follows. Let εi have pdf fε and let g=-f/f. Let W1,W2,,Wn be order statistics for a random sample of size n with the density f.. Define Zi=gWi, then ai=EZri. To define M we need M-1=XTB-AX, where B is an n by n diagonal matrix with Bii=EgWri and A is a symmetric matrix with Aij=covZri,Zrj. In the case of the Normal distribution, the Z1<<Zn are standard Normal order statistics and EgWi=1, for i=1,2,,n.
The analysis can also deal with ties in the data. Two observations are adjudged to be tied if Yi-Yj<tol, where tol is a user-supplied tolerance level.
Various statistics can be found from the analysis:
(a) The score statistic XTa. This statistic is used to test the hypothesis H0:β=0, see (e).
(b) The estimated variance-covariance matrix XTB-AX of the score statistic in (a).
(c) The estimate β^=MXTa.
(d) The estimated variance-covariance matrix M=XTB-AX -1 of the estimate β^.
(e) The χ2 statistic Q=β^TM-1β^=aTXXTB-AX -1XTa used to test H0:β=0. Under H0, Q has an approximate χ2-distribution with p degrees of freedom.
(f) The standard errors Mii 1/2 of the estimates given in (c).
(g) Approximate z-statistics, i.e., Zi=β^i/seβ^i for testing H0:βi=0. For i=1,2,,n, Zi has an approximate N0,1 distribution.
In many situations, more than one sample of observations will be available. In this case we assume the model
hkYk= XkT β+ek,  k=1,2,,ns,  
where ns is the number of samples. In an obvious manner, Yk and Xk are the vector of observations and the design matrix for the kth sample respectively. Note that the arbitrary transformation hk 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 β as β^=Dd, where
D-1=k=1ns XkT Bk-AkXk  
and
d=k= 1ns XkT ak ,  
with ak, Bk and Ak defined as a, B and A above but for the kth sample.
The remaining statistics are calculated as for the one sample case.

4
References

Pettitt A N (1982) Inference for the linear model using a likelihood based on ranks J. Roy. Statist. Soc. Ser. B 44 234–243

5
Arguments

1:     ns – IntegerInput
On entry: the number of samples.
Constraint: ns1.
2:     nvns – Integer arrayInput
On entry: the number of observations in the ith sample, for i=1,2,,ns.
Constraint: nvi1, for i=1,2,,ns.
3:     nsum – IntegerInput
On entry: the total number of observations.
Constraint: nsum= i=1 ns nvi .
4:     ynsum – Real (Kind=nag_wp) arrayInput
On entry: the observations in each sample. Specifically, y k=1 i-1 nvk+j  must contain the jth observation in the ith sample.
5:     ip – IntegerInput
On entry: the number of parameters to be fitted.
Constraint: ip1.
6:     xldxip – Real (Kind=nag_wp) arrayInput
On entry: the design matrices for each sample. Specifically, x k=1 i-1 nvk +j l  must contain the value of the lth explanatory variable for the jth observation in the ith sample.
Constraint: x must not contain a column with all elements equal.
7:     ldx – IntegerInput
On entry: the first dimension of the array x as declared in the (sub)program from which g08raf is called.
Constraint: ldxnsum.
8:     idist – IntegerInput
On entry: the error distribution to be used in the analysis.
idist=1
Normal.
idist=2
Logistic.
idist=3
Extreme value.
idist=4
Double-exponential.
Constraint: 1idist4.
9:     nmax – IntegerInput
On entry: the value of the largest sample size.
Constraint: nmax=max1insnvi and nmax>ip.
10:   tol – Real (Kind=nag_wp)Input
On entry: the tolerance for judging whether two observations are tied. Thus, observations Yi and Yj are adjudged to be tied if Yi-Yj<tol.
Constraint: tol>0.0.
11:   prvrldprvrip – Real (Kind=nag_wp) arrayOutput
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 1ijip, prvrij contains an estimate of the covariance between the ith and jth score statistics. For 1jiip-1, prvri+1j contains an estimate of the covariance between the ith and jth parameter estimates.
12:   ldprvr – IntegerInput
On entry: the first dimension of the array prvr as declared in the (sub)program from which g08raf is called.
Constraint: ldprvrip+1.
13:   iranknmax – Integer arrayOutput
On exit: for the one sample case, irank contains the ranks of the observations.
14:   zinnmax – Real (Kind=nag_wp) arrayOutput
On exit: for the one sample case, zin contains the expected values of the function g. of the order statistics.
15:   etanmax – Real (Kind=nag_wp) arrayOutput
On exit: for the one sample case, eta contains the expected values of the function g. of the order statistics.
16:   vapvecnmax×nmax+1/2 – Real (Kind=nag_wp) arrayOutput
On exit: for the one sample case, vapvec contains the upper triangle of the variance-covariance matrix of the function g. of the order statistics stored column-wise.
17:   parest4×ip+1 – Real (Kind=nag_wp) arrayOutput
On exit: the statistics calculated by the routine.
The first ip components of parest contain the score statistics.
The next ip elements contain the parameter estimates.
parest2×ip+1 contains the value of the χ2 statistic.
The next ip elements of parest contain the standard errors of the parameter estimates.
Finally, the remaining ip elements of parest contain the z-statistics.
18:   worklwork – Real (Kind=nag_wp) arrayWorkspace
19:   lwork – IntegerInput
On entry: the dimension of the array work as declared in the (sub)program from which g08raf is called.
Constraint: lworknmax×ip+1.
20:   iwanmax – Integer arrayWorkspace
21:   ifail – IntegerInput/Output
On entry: ifail must be set to 0, -1​ or ​1. If you are unfamiliar with this argument you should refer to Section 3.4 in How to Use the NAG Library and its Documentation for details.
For environments where it might be inappropriate to halt program execution when an error is detected, the value -1​ or ​1 is recommended. If the output of error messages is undesirable, then the value 1 is recommended. Otherwise, if you are not familiar with this argument, the recommended value is 0. When the value -1​ or ​1 is used it is essential to test the value of ifail on exit.
On exit: ifail=0 unless the routine detects an error or a warning has been flagged (see Section 6).

6
Error Indicators and Warnings

If on entry ifail=0 or -1, explanatory error messages are output on the current error message unit (as defined by x04aaf).
Errors or warnings detected by the routine:
ifail=1
On entry,ns<1,
ortol0.0,
ornmaxip,
orldprvr<ip+1,
orldx<nsum,
ornmaxmax1ins nvi,
ornvi0, for some i, nvi,
ornsumi=1nsnvi,
orip<1,
orlwork<nmax×ip+1.
ifail=2
On entry,idist<1,
oridist>4.
ifail=3
On entry, all the observations are adjudged to be tied. You are advised to check the value supplied for tol.
ifail=4
The matrix XTB-AX is either ill-conditioned or not positive definite. This error should only occur with extreme rankings of the data.
ifail=5
The matrix X has at least one of its columns with all elements equal.
ifail=-99
An unexpected error has been triggered by this routine. Please contact NAG.
See Section 3.9 in How to Use the NAG Library and its Documentation for further information.
ifail=-399
Your licence key may have expired or may not have been installed correctly.
See Section 3.8 in How to Use the NAG Library and its Documentation for further information.
ifail=-999
Dynamic memory allocation failed.
See Section 3.7 in How to Use the NAG Library and its Documentation for further information.

7
Accuracy

The computations are believed to be stable.

8
Parallelism and Performance

g08raf is threaded by NAG for parallel execution in multithreaded implementations of the NAG Library.
g08raf makes calls to BLAS and/or LAPACK routines, which may be threaded within the vendor library used by this implementation. Consult the documentation for the vendor library for further information.
Please consult the X06 Chapter Introduction for information on how to control and interrogate the OpenMP environment used within this routine. Please also consult the Users' Note for your implementation for any additional implementation-specific information.

9
Further Comments

The time taken by g08raf depends on the number of samples, the total number of observations and the number of arguments 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.

10
Example

A program to fit a regression model to a single sample of 20 observations using two explanatory variables. The error distribution will be taken to be logistic.

10.1
Program Text

Program Text (g08rafe.f90)

10.2
Program Data

Program Data (g08rafe.d)

10.3
Program Results

Program Results (g08rafe.r)

© The Numerical Algorithms Group Ltd, Oxford, UK. 2017