In this paper, we propose a modified formulation of the principal components analysis, based on the use of a multivariate Cauchy likelihood instead of the Gaussian likelihood, which has the effect of robustifying the principal components. We present an algorithm to compute these robustified principal components. We additionally derive the relevant influence function of the first component and examine its theoretical properties.
Two new distributions are proposed: the circular projected and the spherical projected Cauchy distributions. A special case of the circular projected Cauchy coincides with the wrapped Cauchy distribution, and for this, a generalization is suggested that offers better fit via the inclusion of an extra parameter. For the spherical case, by imposing two conditions on the scatter matrix we end up with an elliptically symmetric distribution.
Compositional data arise in many real-life applications and versatile methods for properly analyzing this type of data in the regression context are needed. When parametric assumptions do not hold or are difficult to verify, non-parametric regression models can provide a convenient alternative method for prediction. To this end, we consider an extension to the classical k-NN regression, termed a-k-NN regression, that yields a highly flexible non-parametric regression model for compositional data through the use of the a-transformation.
In this paper, we employ a meta-regression analysis approach to synthesize empirical evidence on the average partial effects of eleven adoption determinants that regularly appear in empirical studies examining farmer's adoption behavior worldwide. Our analysis considers a total of 122 studies from the adoption literature using discrete choice models that are published in 24 peer-reviewed journals since 1985, covering farmer's adoption behavior around the world and for a wide variety of agricultural technologies.
This paper develops a novel empirical framework for estimating individual emission levels in a nonpoint source (NPS) pollution problem. For doing so we incorporate into the GME model suggested by Kaplan et al., (2003) a specific theoretical structure describing both crop production technology and nature's residual generating mechanism based on the multiple production relations model suggested by Murty et al, (2012) fitted into a parametric stochastic framework.
We compared maximum likelihood and the k-NN algorithm in the context of discriminant analysis with spherical data.
We present a new model for analyzing compositional data with structural zeros. Inspired by \cite{butler2008} who suggested a model in the presence of zero values in the data we propose a model that treats the zero values in a different manner. Instead of projecting every zero value towards a vertex, we project them onto their corresponding edge and fit a zero-censored multivariate model.
Discriminant analysis for spherical data, or directional data in general, has not been extensively studied, and most papers focus on one distribution, the von Mises-Fisher. In this work, we study more distributions, escaping the rotational symmetry bound of the aforementioned distribution and also include a non parametric classier, the k-NN algorithm.
A folded type model is developed for analyzing compositional data based that provides a new and flexible class of distributions for modeling data defined on the simplex sample space. Despite its rather seemingly complex structure, employment of the EM algorithm guarantees efficient parameter estimation.
Join the notification list of the Department of Economics.