Instances of DensityModel store the estimated densities for each component and the overall (marginal) estimate of the density. The derived class DensityBatchModel additionally stores the density for each batch / component combination (i.e., if there are 3 components and 10 batches, there are 30 estimated densities). The intended use-case of the DensityModel class is to faciliate visualization of the estimated densities (see examples) as well as to provide an estimate of the number of modes in the overall density. If the number of estimated modes is smaller than the number of components of the best-fitting mixture model, post-hoc merging of components may be useful.
The component densities.
The overall (marginal across batches and components) estimate of the density.
A numeric vector providing the estimated modes in the
overall density. The modes are defined by a crude estimate of the
first derivative of the overall density (see
A numeric vector containing the data
A vector providing the k-means clustering of the
component means using the modes as centers. If an object of class
DensityModel is instantiated with
slot takes values 1, ..., K, where K is the number of components.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37
## marginal model truth <- simulateData(N=2500, p=rep(1/3, 3), theta=c(-1, 0, 1), sds=rep(0.1, 3)) dm <- DensityModel(truth) print(dm) dm.merged <- DensityModel(truth, merge=TRUE) print(dm.merged) ## here, because there are 3 distinct modes, specifying merge=TRUE ## does not change the resulting clusters identical(clusters(dm), clusters(dm.merged)) ## These objects can be plotted plot(dm) ## Note that calling plot on a MixtureModel-derived object returns ## a density object as a side-effect of the plotting dm2 <- CNPBayes::plot(truth) identical(dm, dm2) ## batch model k <- 3 nbatch <- 3 means <- matrix(c(-1.2, -1.0, -0.8, -0.2, 0, 0.2, 0.8, 1, 1.2), nbatch, k, byrow=FALSE) sds <- matrix(0.1, nbatch, k) N <- 1500 truth <- simulateBatchData(N=N, batch=rep(letters[1:3], length.out=N), theta=means, sds=sds, p=c(1/5, 1/3, 1-1/3-1/5)) dm <- DensityModel(truth) dm.merged <- DensityModel(truth, merge=TRUE) print(dm) dm2 <- CNPBayes::plot(truth) identical(dm, dm2) ## suppress plotting of the batch-specific densities CNPBayes::plot(dm2, show.batch=FALSE)