Research Articles (Statistics and Actuarial Science)

Permanent URI for this collection

Browse

Recent Submissions

Now showing 1 - 5 of 44
  • Item
    Comparison of zero replacement strategies for compositional data with large numbers of zeros
    (Elsevier B.V., 2021-03) Lubbe, Sugnet; Filzmoser, Peter; Templ, Matthias
    Modern applications in chemometrics and bioinformatics result in compositional data sets with a high proportion of zeros. An example are microbiome data, where zeros refer to measurements below the detection limit of one count. When building statistical models, it is important that zeros are replaced by sensible values. Different replacement techniques from compositional data analysis are considered and compared by a simulation study and examples. The comparison also includes a recently proposed method (Templ, 2020) [1] based on deep learning. Detailed insights into the appropriateness of the methods for a problem at hand are provided, and differences in the outcomes of statistical results are discussed.
  • Item
    Does money "buy" tolerance toward damage-causing wildlife?
    (Wiley, 2020-07-26) Kansky, Ruth; Kidd, Martin; Fischer, Joern
    The Kavango-Zambezi Transfrontier Conservation Area supports large-scale migrations of wildlife that occur in a mixed agri-conservation landscape in five Southern African countries. Human–Wildlife Conflict is a key challenge and understanding the drivers of communities' willingness to coexist with wildlife is thus critical. Community based natural resource management (CBNRM) is a widely used economic approach to foster human-wildlife coexistence with the assumption that monetary benefits can “buy” tolerance by offsetting the disservices of living with wildlife. We tested this assumption and hypothesized that Namibians would be more tolerant towards wildlife than Zambians because they received higher monetary benefits from wildlife. We used the Wildlife Tolerance Model (WTM) as the framework to define tolerance and identify tolerance drivers. We found Namibians tolerance was higher for lion, elephant and hyena but not for kudu and baboon. After controlling for confounding variables of the WTM that could potentially explain differences in tolerance, contrary to expectation, the monetary benefits did not account for higher Namibian tolerance. Instead, only nonmonetary benefits explained the higher tolerance. We used crowding theory to explain this finding, proposing that CBNRM in Namibia and the monetary benefits from the program “crowd in” intrinsic motivation to appreciate and tolerate wildlife.
  • Item
    Variable contribution identification and visualization in multivariate statistical process monitoring
    (Elsevier, 2020-01) Rossouw, R. F.; Coetzer, R. L. J.; Le Roux, N. J.
    Multivariate statistical process monitoring (MSPM) has received book-length treatments and wide spread application in industry. In MSPM, multivariate data analysis techniques such as principal component analysis (PCA) are commonly employed to project the (possibly many) process variables onto a lower dimensional space where they are jointly monitored given a historical or specified reference set that is within statistical control. In this paper, PCA and biplots are employed together in an innovative way to develop an efficient multivariate process monitoring methodology for variable contribution identification and visualization. The methodology is applied to a commercial coal gasification production facility with multiple parallel production processes. More specifically, it is shown how the methodology is used to specify the optimal principal component combinations and biplot axes for visualization and interpretation of process performance, and for the identification of the critical variables responsible for performance deviations, which yielded direct benefits for the commercial production facility.
  • Item
    Involvement of the spinal cord in primary mitochondrial disorders : a neuroimaging mimicker of inflammation and ischemia in children
    (American Journal of Neuroradiology, 2021-02) Alves, C. A. P. F.; Goldstein, A.; Teixeira, S. R.; Martin-Saavedra, J. S.; De Barcelos, I. P.; Fadda, G.; Caschera, L.; Kidd, M.; Gonçalves, F. G.; McCormick, E. M.; Falk, M. J.; Zolkipli-Cunningham, Z.; Vossough, A.; Zuccoli, G.
    Background and purpose: Little is known about imaging features of spinal cord lesions in mitochondrial disorders. The aim of this research was to assess the frequency, imaging features, and pathogenic variants causing primary mitochondrial disease in children with spinal cord lesions. Materials and methods: This retrospective analysis included patients seen at Children's Hospital of Philadelphia between 2000 and 2019 who had a confirmed diagnosis of a primary (genetic-based) mitochondrial disease and available MR imaging of the spine. The MR imaging included at least both sagittal and axial fast spin-echo T2-weighted images. Spine images were independently reviewed by 2 neuroradiologists. Location and imaging features of spinal cord lesions were correlated and tested using the Fisher exact test. Results: Of 119 children with primary mitochondrial disease in whom MR imaging was available, only 33 of 119 (28%) had available spine imaging for reanalysis. Nineteen of these 33 individuals (58%) had evidence of spinal cord lesions. Two main patterns of spinal cord lesions were identified: group A (12/19; 63%) had white ± gray matter involvement, and group B (7/19; 37%) had isolated gray matter involvement. Group A spinal cord lesions were similar to those seen in patients with neuromyelitis optica spectrum disorder, multiple sclerosis, anti-myelin oligodendrocyte glycoprotein-IgG antibody disease, and leukoencephalopathy with brain stem and spinal cord involvement and lactate elevation. Group B patients had spinal cord findings similar to those that occur with ischemia and viral infections. Significant associations were seen between the pattern of lesions (group A versus group B) and the location of lesions in cervical versus thoracolumbar segments, respectively (P < .01). Conclusions: Spinal cord lesions are frequently observed in children with primary mitochondrial disease and may mimic more common causes such as demyelination and ischemia.
  • Item
    On the convergence of gaussian belief propagation with nodes of arbitrary size
    (Journal of Machine Learning Research, 2019) Kamper, Francois; Steel, Sarel J.; Du Preez, Johan A.
    This paper is concerned with a multivariate extension of Gaussian message passing applied to pairwise Markov graphs (MGs). Gaussian message passing applied to pairwise MGs is often labeled Gaussian belief propagation (GaBP) and can be used to approximate the marginal of each variable contained in the pairwise MG. We propose a multivariate extension of GaBP (we label this GaBP-m) that can be used to estimate higher-dimensional marginals. Beyond the ability to estimate higher-dimensional marginals, GaBP-m exhibits better convergence behavior than GaBP, and can also provide more accurate univariate marginals. The theoretical results of this paper are based on an extension of the computation tree analysis conducted on univariate nodes to the multivariate case. The main contribution of this paper is the development of a convergence condition for GaBP-m that moves beyond the walk-summability of the precision matrix. Based on this convergence condition, we derived an upper bound for the number of iterations required for convergence of the GaBP-m algorithm. An upper bound on the dissimilarity between the approximate and exact marginal covariance matrices was established. We argue that GaBP-m is robust towards a certain change in variables, a property not shared by iterative solvers of linear systems, such as the conjugate gradient (CG) and preconditioned conjugate gradient (PCG) methods. The advantages of using GaBP-m over GaBP are also illustrated empirically.