Welcome to GeoModels
The GeoModels package provides a set of procedures for simulation, estimation and prediction of spatio-temporal random fields.
The main features of the package are:
The type of data that can be modeled are:
- spatial data
- space time data with spatial location sites possibly changing over time
- bivariate spatial data with (possibly different) spatial location sites
Data can be defined on euclidean space or on a sphere of arbitrary radius
The random fields can have the following marginal distributions:
- Continuous distribution supported on the whole real line:
- Student T
- Positive continuous distributions:
- Discrete distributions:
- Negative binomial
- Circular distributions:
Parametric models for both regression and dependence analysis through covariance models
Parametric (bivariate) spatial and spatiotemporal covariance models, including Matern, Generalized Wendland, Gneiting model, bivariate Matern
- Composite likelihood based on pairs (optional parallel computation with OpenCL)
- Full likelihood (when feasible)
Optimal (local) linear prediction (kriging)
Tutorials and Examples
|Gaussian||Analysis of spatial data using a flexible compactly supported covariance model||Link||Wend_Matern.R|
|Gaussian||Analysis of global spatial data on the planet Earth using Gaussian random fields||Link||gausphere.R|
|Skew-Gaussian||Analysis of asymmetric spatial data using skew-Gaussian random fields||Link||skewg.R|
|t||Analysis of heavy tails spatial data using t random fields||Link||t.R|
|Tukey-h||Analysis of heavy tails spatial data using tukey-h random fields||Link||tukey.R|
|Weibull||Analysis of positive spatial data using Weibull random fields||Link||weibull.R|
|Log-Gaussian||Analysis of positive spatial data using Log-Gaussian random fields||Link||loggausssian.R|
|Poisson||Analysis of spatial count data using Poisson random fields||Link||poisson.R|
|Binomial||Analysis of spatial discrete data using Binomial random fields||Link||binomial.R|
|Gaussian||Analysis of spatio-temporal data using Gaussian random fields||Link||ST.R|
|Gaussian||Analysis of spatio-temporal data with spatial locations changing over time||Link||ST_dyn.R|
Spatial bivariate Data
|Gaussian||Analysis of bivariate spatial data using bivariate Gaussian random fields||Link||biv.R|
Real data analysis
|Skew-Gaussian||Analysis of spatial precipitation anomalies using Gaussian and skew Gaussian random fields||Link||anomalies.R|
|t||Analysis of maximum temperature of Australia using t random fields||Link||austr.R|
Resources and Download
Latest binaries and sources for GeoModels are availables from GitHub repository:
- GeoModels Package sources for any platform
- GeoModels OpenCL for OX only
- Reference Manual
- Bugs report
Option 1: CRAN version
Our cross-platform version can be found at CRAN:
GeoModels package. You can install it with:
Option 2: Developer
We currently are loaded in Github only. This means that for
GeoModels installation you will need to previously install
devtools package if you do not have it installed yet:
devtools lets you install packages from github since they need to be installed from source code.
We have developed two GeoModels version, one standard version and one that uses the
OpenCL framework for parallel computing. The standard version can be installed in any operating system: Windows, OSX and Linux,
and you are good to go.
A word of caution though. In Windows, make sure
RTools is installed in order to build packages.
Rtools must be compatible with your current
R version. This usually happens with
Rtools latest release (which is
Rtools.35.exe today), you can download
here. Also, we have lately received some issues regarding
devtools since the last release has a slight problem. To avoid this, install version 1.13.6 that you can find
GeoModelsversion is currently supported for OSX (Sierra and Mojave). All OpenCL code has been tested on
Intel(R) Core(TM) i7-4980HQ CPU @ 2.80GHz. We are currently developing the OpenCL
GeoModelsversion to work in any operating system and debbuging it for other graphics cards. It is installed with this code:
gpuR packageand try an example from that package so that OpenCL headers are tested. Installation instructions are in this
- Morales-Navarrete D., Bevilacqua M., Caamaño C., Castro L.M. (2022) Modelling Point Referenced Spatial Count Data: A Poisson Process Approach. Journal of the American Statistical Association. Accepted 10.1080/01621459.2022.2140053
- Caamaño-Carrillo C, Bevilacqua M, López C, Morales-Oñate V (2022) Nearest neighbours weighted composite likelihood based on pairs for (non-)Gaussian massive spatial data with an application to Tukey-hh random fields estimation Research Square. 10.21203/rs.3.rs-2073895/v1
- Blasi F., Caamaño C., Bevilacqua M., Furrer R. (2022) A selective view of climatological data and likelihood estimation Spatial Statistics . 10.1016/j.spasta.2022.100596
- Bevilacqua M., Caamaño C., Porcu E. (2022) Unifying Compactly Supported and Matern Covariance Functions in Spatial Statistics Journal of Multivariate Analysis. 10.1016/j.jmva.2022.104949
- Bevilacqua M., Caamaño C., Arellano-Valle R. B., Camilo Gomez C. (2022) A class of random fields with two-piece marginal distributions for modeling point-referenced data with spatial outliers. Test. DOI: 10.1007/s11749-021-00797-5
- Bevilacqua, M., C. Caamano, and C. Gaetan (2020). On modelling positive continuous data with spatio-temporal dependence. Environmetrics. 31(7).
- Bevilacqua, M., C. Caamano, R. B. A. Valle, and V. Morales-Oñate (2019). On Spatial (Skew) t Processes and Applications. Scandinavian Journal of Statistics. 48, 212– 245.
- Bevilacqua, M., T. Faouzi, R. Furrer, and E. Porcu (2019). Estimation and prediction using generalized Wendland covariance function under fixed domain asymptotics. The Annals of Statistics. 47(2) 828–856 Annals of Statistics.
- Alegria, A., S. Caro, M. Bevilacqua, E. Porcu, and J. Clarke (2017). Estimating covariance functions of multivariate skew-gaussian random fields on the sphere. Spatial Statistics 22, 388 – 402
- Alegria A., Bevilacqua, M., Porcu, E. (2016) Likelihood-based inference for multivariate space-time wrapped-Gaussian fields. Journal of Statistical Computation and Simulation. 86(13) 2583-2597.
- Bevilacqua M., Alegria A., Velandia D.,Porcu E. (2016) Composite likelihood inference for multivariate Gaussian random fields. Journal of Agricultural Biological and Environmental Statistics. 21(3), 448-469
- Bevilacqua , M., Vallejos, R. and Velandia D. (2015). Assessing the significance of the correlation between the components of a bivariate Gaussian random field. Environmetrics. 26(8), 545-556.
- Bevilacqua, M. and C. Gaetan (2015). Comparing composite likelihood methods based on pairs for spatial Gaussian random fields. Statistics and Computing 25, 877–892.
- Bevilacqua, M., C. Gaetan, J. Mateu, and E. Porcu (2012). Estimating space and space- time covariance functions for large data sets: a weighted composite likelihood approach. Journal of the American Statistical Association 107, 268–280
- Zhang, H. and A. El-Shaarawi (2010). On spatial skew-Gaussian processes and applications. Environmetrics 21(1), 33–47.
Once you have installed
GeoModels, you can have a BibTex citation with
citation("GeoModels") and get:
About the authors
Moreno Bevilacqua is an Associate Professor at the Faculty of Engineering and Science of Adolfo Ibañez University in Viña del Mar (Chile) from August 2020.
He worked at the Statistics Department of University of Valparaiso (Chile) from August 2012.
He has carried out research as: a post-doc at the Department of Statistics, University Ca' Foscari of Venice from May 2008 to December 2010, a research fellow at the University of Bergamo from January 2011 to July 2012.
He received his PhD in Statistics in 2008 and his Degree in Statistics in 2001 from the University of Padua. His main research interests concern theory, methodology and applications in multivariate spatio-temporal statistics.
Víctor received his PhD in Statistics from Universidad de Valparaíso-Chile in 2018. His main research interests concern computational spatial (temporal) geostatistics applications.
Other interests include data analytics in business, machine learning, quantitative economy and philosophy, particularly, Bertrand Russell's political ideas.
Christian Caamaño-Carrillo is an Assistant Professor at the Statistics Department of Universidad del Bío-Bío from 2013. He received his PhD in Statistics from Universidad de Valparaíso in 2018 and his Master's in Statistics in 2012.
His main research interests concern theory, methodology and applications in space-time statistics for non-Gaussian data.