Arnošt Komárek

In this page, several extension packages for the statistical software R and a macro for the statistical package SAS written in the framework of my research are presented.


R is a popular open source software tool for statistical analysis and graphics. More information can be obtained from its webpage. Many extension packages are available from the Comprehensive R Archive Network (CRAN) accessible from the R webpage.

I am author of several R extension packages. All of them are available under the GNU license. Package mixAK contains several methods to analyze normal mixtures and perform a model based clustering. Packages smoothSurv and bayesSurv implement mainly methods for the regression analysis with time-to-event type of data (survival analysis). Package glmmAK is devoted to generalized linear mixed models.

Package mixAK

This package mainly implements MCMC analyzis of (multivariate) normal mixtures. For univariate mixtures, the number of components may be estimated jointly with remaining parameters using reversible jump MCMC. For multivariate mixtures, several criteria (PED, DIC) are implemented to guide the selection of the number of components. The package allows for right-, left-, and interval-censored data. Additionally, model based clustering for longitudinal and otherwise correlated data is implemented here.

The package is described in

The methodology is described and the package used in

This package is available from CRAN.

mixAK on CRAN  

Package smoothSurv

The package implements mainly an accelerated failure time model with an error distribution specified as a penalized normal mixture where the parameters are estimated using a method of penalized maximum-likelihood. Besides more common right-censored data, the package can also be used to analyze interval-censored data.

The methodology is described and the package used in

This package is available from CRAN.

smoothSurv on CRAN  

Package bayesSurv

In the package, several accelerated failure time models with random effects are implemented. The error distribution and/or the distribution of the random effects is modelled either using a classical normal mixture with unspecified number of components or using a penalized normal mixture where the number of components is overspecified. The model parameters are estimated using the Markov chain Monte Carlo method in the Bayesian framework.

The methodology is described and the package used in

This package is available from CRAN.

bayesSurv on CRAN  

Package glmmAK

This package implements maximum-likelihood estimation in the logistic regression with both binary (logit model) and multinomial response (cumulative logit model), and in the Poisson regression (log-linear model). Secondly, Bayesian estimation based on MCMC in the logistic and Poisson regression model with random effects whose distribution is specified as a penalized normal mixture are implemented.

The methodology is described and the package used in

This package is no more available from CRAN.

Source code   Windows binary   Manual      glmmAK on CRAN (archive)  


SAS is a commercial statistical software. Its webpage tells you more. In 2001, I wrote a SAS macro called hetmixed.

Macro hetmixed

The macro can be used to fit the linear mixed model with a normal mixture as a distribution for random effects. The methodology has been described on the following places.
  • Verbeke, G. and Lesaffre, E. (1997). A linear mixed-effects model with heterogeneity in the random-effects population. Journal of the American Statistical Association, 91, 217-221.
  • Verbeke, G. and Molenberghs, G. (2000). Linear Mixed Models for Longitudinal Data. New York: Springer-Verlag, pp. 169-188.

The macro itself is documented in my MSc. thesis (Komárek, 2001) from which a manual has been derived.

The macro is freely available to everyone and I guess it still works. You can use it but, please, do not ask any help from me anymore. The macro was written more than 10 years ago and that is also the time I used it actively for the last time.

Source code   Manual   Example code  


View My Stats