Perform nonparametric Bayesian analysis using Dirichlet processes without the need to program the inference algorithms. Utilise included pre-built models or specify custom models and allow the 'dirichletprocess' package to handle the Markov chain Monte Carlo sampling. Our Dirichlet process objects can act as building blocks for a variety of statistical models including and not limited to: density estimation, clustering and prior distributions in hierarchical models. See Teh, Y. W. (2011) < https://www.stats.ox.ac.uk/~teh/research/npbayes/Teh2010a.pdf>, among many other sources.
The dirichletprocess package provides tools for you to build custom Dirichlet process mixture models. You can use the pre-built Normal/Weibull/Beta distributions or create your own following the instructions in the vignette. In as little as four lines of code you can be modelling your data nonparametrically.
You can install the stable release of dirichletprocess from CRAN:
You can also install the development build of dirichletprocess from github with:
For a full guide to the package and its capabilities please consult the vignette:
browseVignettes(package = "dirichletprocess")
Dirichlet processes can be used for nonparametric density estimation.
faithfulTransformed <- faithful$waiting - mean(faithful$waiting)faithfulTransformed <- faithfulTransformed/sd(faithful$waiting)dp <- DirichletProcessGaussian(faithfulTransformed)dp <- Fit(dp, 100, progressBar = FALSE)plot(dp)
data.frame(Weight=dp$weights, Mean=c(dp$clusterParameters[]), SD=c(dp$clusterParameters[]))#> Weight Mean SD#> 1 0.371323529 -1.1756510 -1.1756510#> 2 0.625000000 0.6597522 0.6597522#> 3 0.003676471 0.1061095 0.1061095
Dirichlet processes can also be used to cluster data based on their common distribution parameters.
faithfulTrans <- as.matrix(apply(faithful, 2, function(x) (x-mean(x))/sd(x)))dpCluster <- DirichletProcessMvnormal(faithfulTrans)dpCluster <- Fit(dpCluster, 1000, progressBar = FALSE)
To plot the results we take the cluster labels contained in the
dp object and assign them a colour
For more detailed explanations and examples see the vignette.
plot. Multivariate Gaussian models can now be plotted.
NEWS.mdfile to track changes to the package.