The routine may be called by the names g02gcf or nagf_correg_glm_poisson.
A generalized linear model with Poisson errors consists of the following elements:
(a)a set of observations, , from a Poisson distribution:
(b), a set of independent variables for each observation, .
(c)a linear model:
(d)a link between the linear predictor, , and the mean of the distribution, , . The possible link functions are:
(i)exponent link: , for a constant ,
(ii)identity link: ,
(iii)log link: ,
(iv)square root link: ,
(v)reciprocal link: .
(e)a measure of fit, the deviance:
The linear parameters are estimated by iterative weighted least squares. An adjusted dependent variable, , is formed:
and a working weight, ,
At each iteration an approximation to the estimate of , , is found by the weighted least squares regression of on with weights .
g02gcf finds a decomposition of , i.e., where is a triangular matrix and is an column orthogonal matrix.
If is of full rank, then is the solution to:
If is not of full rank a solution is obtained by means of a singular value decomposition (SVD) of .
where is a diagonal matrix with nonzero diagonal elements, being the rank of and .
This gives the solution
being the first columns of , i.e., .
The iterations are continued until there is only a small change in the deviance.
The initial values for the algorithm are obtained by taking
The fit of the model can be assessed by examining and testing the deviance, in particular by comparing the difference in deviance between nested models, i.e., when one model is a sub-model of the other. The difference in deviance between two nested models has, asymptotically, a -distribution with degrees of freedom given by the difference in the degrees of freedom associated with the two deviances.
The parameters estimates, , are asymptotically Normally distributed with variance-covariance matrix
in the full rank case, otherwise
The residuals and influence statistics can also be examined.
The estimated linear predictor , can be written as for an matrix . The th diagonal elements of , , give a measure of the influence of the th values of the independent variables on the fitted regression model. These are known as leverages.
The fitted values are given by .
g02gcf also computes the deviance residuals, :
An option allows prior weights to be used with the model.
In many linear regression models the first term is taken as a mean term or an intercept, i.e., , for . This is provided as an option.
Often only some of the possible independent variables are included in a model; the facility to select variables to be included in the model is provided.
If part of the linear predictor can be represented by a variables with a known coefficient then this can be included in the model by using an offset, :
If the model is not of full rank the solution given will be only one of the possible solutions. Other estimates may be obtained by applying constraints to the parameters. These solutions can be obtained by using g02gkf after using g02gcf. Only certain linear combinations of the parameters will have unique estimates, these are known as estimable functions, these can be estimated and tested using g02gnf.
Details of the SVD are made available in the form of the matrix :
On exit: the degrees of freedom asociated with the deviance for the fitted model.
16: – Real (Kind=nag_wp) arrayOutput
On exit: the estimates of the parameters of the generalized linear model, .
If , the first element of b will contain the estimate of the mean parameter and will contain the coefficient of the variable contained in column of , where is the th positive value in the array isx.
If , will contain the coefficient of the variable contained in column of , where is the th positive value in the array isx.
17: – IntegerOutput
On exit: the rank of the independent variables.
If the model is of full rank, .
If the model is not of full rank, irank is an estimate of the rank of the independent variables. irank is calculated as the number of singular values greater that (largest singular value). It is possible for the SVD to be carried out but for irank to be returned as ip.
18: – Real (Kind=nag_wp) arrayOutput
On exit: the standard errors of the linear parameters.
contains the standard error of the parameter estimate in , for .
19: – Real (Kind=nag_wp) arrayOutput
On exit: the upper triangular part of the variance-covariance matrix of the ip parameter estimates given in b. They are stored packed by column, i.e., the covariance between the parameter estimate given in and the parameter estimate given in , , is stored in .
, for , must contain the offset values . All other values need not be set.
On exit: auxiliary information on the fitted model.
contains the linear predictor value,
, for .
contains the fitted value,
, for .
contains the variance standardization,
, for .
contains the square root of the working weight,
, for .
contains the deviance residual,
, for .
contains the leverage,
, for .
contains the offset,
, for . If , all values will be zero.
for , contains the results of the decomposition or the singular value decomposition.
If the model is not of full rank, i.e., , the first ip rows of columns to contain the matrix.
21: – IntegerInput
On entry: the first dimension of the array v as declared in the (sub)program from which g02gcf is called.
22: – Real (Kind=nag_wp)Input
On entry: indicates the accuracy required for the fit of the model.
The iterative weighted least squares procedure is deemed to have converged if the absolute change in deviance between iterations is less than . This is approximately an absolute precision if the deviance is small and a relative precision if the deviance is large.
If , the routine will use instead.
23: – IntegerInput
On entry: the maximum number of iterations for the iterative weighted least squares.
If , a default value of is used.
24: – IntegerInput
On entry: indicates if the printing of information on the iterations is required.
There is no printing.
Every iprint iteration, the following are printed:
the current estimates;
and if the weighted least squares equations are singular, then this is indicated.
When printing occurs the output is directed to the current advisory message unit (see x04abf).
25: – Real (Kind=nag_wp)Input
On entry: the value of eps is used to decide if the independent variables are of full rank and, if not, what is the rank of the independent variables. The smaller the value of eps the stricter the criterion for selecting the singular value decomposition.
If , the routine will use machine precision instead.
26: – Real (Kind=nag_wp) arrayWorkspace
27: – IntegerInput/Output
On entry: ifail must be set to , or to set behaviour on detection of an error; these values have no effect when no error is detected.
A value of causes the printing of an error message and program execution will be halted; otherwise program execution continues. A value of means that an error message is printed while a value of means that it is not.
If halting is not appropriate, the value or is recommended. If message printing is undesirable, then the value is recommended. Otherwise, the value is recommended since useful values can be provided in some output arguments even when on exit. When the value or is used it is essential to test the value of ifail on exit.
On exit: unless the routine detects an error or a warning has been flagged (see Section 6).
6Error Indicators and Warnings
If on entry or , explanatory error messages are output on the current error message unit (as defined by x04aaf).
Errors or warnings detected by the routine:
Note: in some cases g02gcf may return useful information.
On entry, ip incompatible with number of nonzero values of isx: .
On entry, .
Constraint: , for .
On entry, .
Constraint: , for .
A fitted value is at the boundary, i.e., . This may occur if there are values of and the model is too complex for the data. The model should be reformulated with, perhaps, some observations dropped.
SVD solution failed to converge.
The iterative weighted least squares has failed to converge in maxit (or default ) iterations. The value of maxit could be increased but it may be advantageous to examine the convergence using the iprint option. This may indicate that the convergence is slow because the solution is at a boundary in which case it may be better to reformulate the model.
The rank of the model has changed during the weighted least squares iterations. The estimate for returned may be reasonable, but you should check how the deviance has changed during iterations.
The degrees of freedom for error are . A saturated model has been fitted.
An unexpected error has been triggered by this routine. Please
See Section 7 in the Introduction to the NAG Library FL Interface for further information.
Your licence key may have expired or may not have been installed correctly.
See Section 8 in the Introduction to the NAG Library FL Interface for further information.
Dynamic memory allocation failed.
See Section 9 in the Introduction to the NAG Library FL Interface for further information.
The accuracy depends on the value of tol as described in Section 5. As the deviance is a function of the accuracy of the will only be a function of tol. tol should, therefore, be set smaller than the accuracy required for .
8Parallelism and Performance
Background information to multithreading can be found in the Multithreading documentation.
g02gcf is threaded by NAG for parallel execution in multithreaded implementations of the NAG Library.
g02gcf 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.
A contingency table given by Plackett (1974) is analysed by fitting terms for rows and columns. The table is: