84 research outputs found
High-Dimensional Graphical Model Search with the gRapHD R Package
No full textThis paper presents the R package gRapHD for efficient selection of high-dimensional undirected graphical models. The package provides tools for selecting trees, forests, and decomposable models minimizing information criteria such as AIC or BIC, and for displaying the independence graphs of the models. It has also some useful tools for analysing graphical structures. It supports the use of discrete, continuous, or both types of variables.
On the Efficient Score Function for some Semiparametric Location-Scale Models
No full textThe paper studies some semiparametric extensions of the location-scale models. The efficient score function for these models is calculated using a technique based on expansions in orthogonal polynomial of the parametric partial score function. As an auxiliary result a sufficient condition involving the Laplace transform is given for having the class of polynomials dense in L 2 . In several of the examples considered, the efficient score function depends on the nuisance parameter; however, this dependence is only through an intermediate finite dimensional parameter. In some examples the efficient score function essentially does not depend on the nuisance parameter, thus implying optimality of these estimating functions. Department of Theoretical Statistics, University of Aarhus, and Department of Biometry and Informatics, Foulum Research Center, Danish Ministry of Agriculture and Fisheries. 1 Contents 1 Introduction 3 2 Preliminaries 4 3 Semiparametric Location-Scale Models 7 4..
On the Bias of the Score Function of Finite Mixture Models
No full textWe characterise the unbiasedness of the score function, viewed as an
inference function for a class of finite mixture models. The models studied
represent the situation where there is a stratification of the observations in
a finite number of groups. We show that, under mild regularity conditions, the
score function for estimating the parameters identifying each group's
distribution is unbiased. We also show that if one introduces a mixture in the
scenario described above so that for some observations, it is only known that
they belong to some of the groups with a probability not in , then
the score function becomes biased. We argue then that under further mild
regularity, the maximum likelihood estimate is not consistent. The results
above are extended to regular models containing arbitrary nuisance parameters,
including semiparametric models.Comment: 7 page
Fungi in Danish soils under organic and conventional farming
No full textA multi-soil study was conducted in Denmark including 29 sites, 8 classified as ‘Organic’, 11 as ‘Conventional with manure and synthetic fertilisers’ and 10 as ‘Conventional with synthetic fertilisers’. The variability of fungal abundance within the three farming systems and the long-term effects of different farming systems on fungal propagules in soil were evaluated.
Fungal abundance showed large variations within all three farming systems and this variability reduced the possibility to obtain general conclusions on fungal composition in soils under different farming systems. This was illustrated by the results on total propagule numbers of filamentous fungi and yeasts. Penicillium spp. and Gliocladium roseum were more abundant under organic than conventional farming, while Trichoderma spp. were most abundant in conventionally farmed soils with synthetic fertilisers. These results were not altered after adjusting for possible differences in basic soil properties like total-C and N, extractable P, CEC, base saturation and soil density. The paper discusses whether the differences in fungal abundance are characteristics of a farming system itself or associated with certain management factors being more prevalent in one farming system than the other
Selecting high-dimensional mixed graphical models using minimal AIC or BIC forests
No full textAbstract Background Chow and Liu showed that the maximum likelihood tree for multivariate discrete distributions may be found using a maximum weight spanning tree algorithm, for example Kruskal's algorithm. The efficiency of the algorithm makes it tractable for high-dimensional problems. Results We extend Chow and Liu's approach in two ways: first, to find the forest optimizing a penalized likelihood criterion, for example AIC or BIC, and second, to handle data with both discrete and Gaussian variables. We apply the approach to three datasets: two from gene expression studies and the third from a genetics of gene expression study. The minimal BIC forest supplements a conventional analysis of differential expression by providing a tentative network for the differentially expressed genes. In the genetics of gene expression context the method identifies a network approximating the joint distribution of the DNA markers and the gene expression levels. Conclusions The approach is generally useful as a preliminary step towards understanding the overall dependence structure of high-dimensional discrete and/or continuous data. Trees and forests are unrealistically simple models for biological systems, but can provide useful insights. Uses include the following: identification of distinct connected components, which can be analysed separately (dimension reduction); identification of neighbourhoods for more detailed analyses; as initial models for search algorithms with a larger search space, for example decomposable models or Bayesian networks; and identification of interesting features, such as hub nodes.</p
High-Dimensional Graphical Model Search with the gRapHD R Package
No full textThis paper presents the R package gRapHD for efficient selection of high-dimensional undirected graphical models. The package provides tools for selecting trees, forests, and decomposable models minimizing information criteria such as AIC or BIC, and for displaying the independence graphs of the models. It has also some useful tools for analysing graphical structures. It supports the use of discrete, continuous, or both types of variables
- …
