A tunable radial basis function model for nonlinear system identification using particle swarm optimisation

A tunable radial basis function (RBF) network model is proposed for nonlinear system identification using particle swarm optimisation (PSO). At each stage of orthogonal forward regression (OFR) model construction, PSO optimises one RBF unit's centre vector and diagonal covariance matrix by minimising the leave-one-out (LOO) mean square error (MSE). This PSO aided OFR automatically determines how many tunable RBF nodes are sufficient for modelling. Compared with the-state-of-the-art local regularisation assisted orthogonal least squares algorithm based on the LOO MSE criterion for constructing fixed-node RBF network models, the PSO tuned RBF model construction produces more parsimonious RBF models with better generalisation performance and is computationally more efficient.

Nonlinear system identification using particle swarm optimisation tuned radial basis function models

A novel particle swarm optimisation (PSO) tuned radial basis function (RBF) network model is proposed for identification of non-linear systems. At each stage of orthogonal forward regression (OFR) model construction process, PSO is adopted to tune one RBF unit's centre vector and diagonal covariance matrix by minimising the leave-one-out (LOO) mean square error (MSE). This PSO aided OFR automatically determines how many tunable RBF nodes are sufficient for modelling. Compared with the-state-of-the-art local regularisation assisted orthogonal least squares algorithm based on the LOO MSE criterion for constructing fixed-node RBF network models, the PSO tuned RBF model construction produces more parsimonious RBF models with better generalisation performance and is often more efficient in model construction. The effectiveness of the proposed PSO aided OFR algorithm for constructing tunable node RBF models is demonstrated using three real data sets.

Sparse multioutput radial basis function network construction using combined locally regularised orthogonal least square and D-optimality experimental design

A construction algorithm for multioutput radial basis function (RBF) network modelling is introduced by combining a locally regularised orthogonal least squares (LROLS) model selection with a D-optimality experimental design. The proposed algorithm aims to achieve maximised model robustness and sparsity via two effective and complementary approaches. The LROLS method alone is capable of producing a very parsimonious RBF network model with excellent generalisation performance. The D-optimality design criterion enhances the model efficiency and robustness. A further advantage of the combined approach is that the user only needs to specify a weighting for the D-optimality cost in the combined RBF model selecting criterion and the entire model construction procedure becomes automatic. The value of this weighting does not influence the model selection procedure critically and it can be chosen with ease from a wide range of values.

Data based constructive identification - overcoming the curse of dimensionality

The modelling of a nonlinear stochastic dynamical processes from data involves solving the problems of data gathering, preprocessing, model architecture selection, learning or adaptation, parametric evaluation and model validation. For a given model architecture such as associative memory networks, a common problem in non-linear modelling is the problem of "the curse of dimensionality". A series of complementary data based constructive identification schemes, mainly based on but not limited to an operating point dependent fuzzy models, are introduced in this paper with the aim to overcome the curse of dimensionality. These include (i) a mixture of experts algorithm based on a forward constrained regression algorithm; (ii) an inherent parsimonious delaunay input space partition based piecewise local lineal modelling concept; (iii) a neurofuzzy model constructive approach based on forward orthogonal least squares and optimal experimental design and finally (iv) the neurofuzzy model construction algorithm based on basis functions that are Bézier Bernstein polynomial functions and the additive decomposition. Illustrative examples demonstrate their applicability, showing that the final major hurdle in data based modelling has almost been removed.

A critical review of methodologies for measuring rental depreciation applied to UK commercial real estate

Real estate depreciation continues to be a critical issue for investors and the appraisal profession in the UK in the 1990s. Depreciation-sensitive cash flow models have been developed, but there is a real need to develop further empirical methodologies to determine rental depreciation rates for input into these models. Although building quality has been found to be an important explanatory variable in depreciation it is very difficult to incorporate it into such models or to analyse it retrospectively. It is essential to examine previous depreciation research from real estate and economics in the USA and UK to understand the issues in constructing a valid and pragmatic way of calculating rental depreciation. Distinguishing between 'depreciation' and 'obsolescence' is important, and the pattern of depreciation in any study can be influenced by such factors as the type (longitudinal or crosssectional) and timing of the study, and the market state. Longitudinal studies can analyse change more directly than cross-sectional studies. Any methodology for calculating rental depreciation rate should be formulated in the context of such issues as 'censored sample bias', 'lemons' and 'filtering', which have been highlighted in key US literature from the field of economic depreciation. Property depreciation studies in the UK have tended to overlook this literature, however. Although data limitations and constraints reduce the ability of empirical property depreciation work in the UK to consider these issues fully, 'averaging' techniques and ordinary least squares (OLS) regression can both provide a consistent way of calculating rental depreciation rates within a 'cohort' framework.

Differentiated effects of formal and informal institutional distance between countries 2 on the environmental performance of multinational enterprises

This research examines the influence of environmental institutional distance between home and host countries on the standardization of environmental performance among multinational enterprises using ordinary least-squares (OLS) regression techniques and a sample of 128 multinationals from high-polluting industries. The paper examines the environmental institutional distance of countries using the concepts of formal and informal institutional distances. The results show that whereas a high formal environmental distance between home and host countries leads multinational enterprises to achieve a different level of environmental performance according to each country's legal requirements, a high informal environmental distance encourages these firms to unify their environmental performance independently of the countries in which their units are based. The study also discusses the implications for academia, managers, and policy makers.

Robust inference in an heteroscedastic measurement error model

In this paper we deal with robust inference in heteroscedastic measurement error models Rather than the normal distribution we postulate a Student t distribution for the observed variables Maximum likelihood estimates are computed numerically Consistent estimation of the asymptotic covariance matrices of the maximum likelihood and generalized least squares estimators is also discussed Three test statistics are proposed for testing hypotheses of interest with the asymptotic chi-square distribution which guarantees correct asymptotic significance levels Results of simulations and an application to a real data set are also reported (C) 2009 The Korean Statistical Society Published by Elsevier B V All rights reserved

Estimating the effect of discretionary spending on corruption: evidence from Brazilian municipalities

O presente artigo estuda a relação entre corrupção e discricionariedade do gasto público ao responder a seguinte pergunta: regras de licitação mais rígidas, uma proxy para discricionariedade, resultam em menor prevalência de corrupção nos municípios brasileiros? A estratégia empírica é uma aproximação de regressões em dois estágios (2SLS) estimadas localmente em cada transição de regras de licitação, cuja fonte de dados de corrupção é o Programa de Fiscalização por Sorteio da CGU e os dados sobre discricionariedade são derivados da Lei 8.666/93, responsável por regular os processos de compras e construção civil em todas as esferas de governo. Os resultados mostram, entretanto, que menor discricionariedade está relacionada com maior corrupção para quase todos os cortes impostos pela lei de licitações.

Determinantes da remuneração do spread de certificados de recebíveis imobiliários no mercado brasileiro

Este estudo tem por objetivo identificar as principais variáveis que afetam o spread de Certificados de Recebíveis Imobiliários (CRI) no mercado nacional no momento da emissão do título. Dentre as principais variáveis estudadas estão o volume de emissão, prazo dos títulos, existência de rating, tipo de emissão, dentre outros. Para o trabalho foi utilizada uma base de dados de acompanhamento do mercado realizado pela ANBIMA e contou com 535 emissões, realizadas entre 2005 e 2015, de CRI indexados ao IGP, IPCA, DI e TR, além de títulos pré-fixados. Foram elaboradas cinco regressões pelo método dos Mínimos Quadrados Ordinários (MQO) stepwise sendo que as amostras foram separadas de acordo com o tipo de indexador (IGP, IPCA, DI e TR) e uma que englobou todas as emissões. Os resultados se mostraram distintos para cada regressão realizada e as variáveis explicativas do spread que se mostraram significativas na maioria das regressões foram o volume de emissão e a variável representativa da percepção internacional do cenário brasileiro. A evolução apresentada por este estudo com relação aos trabalhos disponíveis na literatura é de expandir as análises realizadas com outros títulos mobiliários, como Debentures e Fundos de Investimento em Direito Creditórios, para os CRIs.

Fundos de investimento em direitos creditórios: determinantes da taxa de captação dos FIDCs

O mercado de securitização através de Fundos de Investimento em Direitos Creditórios (FIDCs) no Brasil ainda não é tão desenvolvido quando comparado aos mercados internacionais. O desenvolvimento desse mercado facultaria a bancos e empresas acesso a fontes de financiamento além da sua geração de caixa ou da emissão de novas ações. Bancos e empresas devem manter a busca por formas alternativas de financiamento, obtendo, assim, diversificação de suas fontes de recursos. O objetivo desse trabalho é avaliar como as características dos FIDCs afetam a taxa de captação de suas cotas sênior. Para isso foram avaliadas 151 séries de cotas, emitidas entre 2002 e 2014. Na análise foram usadas análises multivariadas através da regressão linear múltipla, usando o método de Mínimos Quadrados Ordinários (MQO), para avaliar o efeito das variáveis sobre o spread das cotas. Os resultados apontaram que o volume de emissão afeta a forma como outras variáveis afetam o spread. Os fundos com tamanho de oferta maior têm seu spread influenciado por características de seus cedentes e por características de sua estruturação, enquanto que fundos menores são apreçados por negociações entre emissor e investidores.

Determinantes da Estrutura de Capital de Companhias Abertas Brasileiras Não-Financeiras: Um Estudo Empírico com Dados em Painel

When a company desires to invest in a project, it must obtain resources needed to make the investment. The alternatives are using firm s internal resources or obtain external resources through contracts of debt and issuance of shares. Decisions involving the composition of internal resources, debt and shares in the total resources used to finance the activities of a company related to the choice of its capital structure. Although there are studies in the area of finance on the debt determinants of firms, the issue of capital structure is still controversial. This work sought to identify the predominant factors that determine the capital structure of Brazilian share capital, non-financial firms. This work was used a quantitative approach, with application of the statistical technique of multiple linear regression on data in panel. Estimates were made by the method of ordinary least squares with model of fixed effects. About 116 companies were selected to participate in this research. The period considered is from 2003 to 2007. The variables and hypotheses tested in this study were built based on theories of capital structure and in empirical researches. Results indicate that the variables, such as risk, size, and composition of assets and firms growth influence their indebtedness. The profitability variable was not relevant to the composition of indebtedness of the companies analyzed. However, analyzing only the long-term debt, comes to the conclusion that the relevant variables are the size of firms and, especially, the composition of its assets (tangibility).This sense, the smaller the size of the undertaking or the greater the representation of fixed assets in total assets, the greater its propensity to long-term debt. Furthermore, this research could not identify a predominant theory to explain the capital structure of Brazilian

O argumento de inclusão enquanto política de acesso à universidade pública

There are a great number of evidences showing that education is extremely important in many economic and social dimensions. In Brazil, education is a right guaranteed by the Federal Constitution; however, in the Brazilian legislation the right to the three stages of basic education: Kindergarten, Elementary and High School is better promoted and supported than the right to education at College level. According to educational census data (INEP, 2009), 78% of all enrolments in College education are in private schools, while the reverse is found in High School: 84% of all matriculations are in public schools, which shows a contradiction in the admission into the universities. The Brazilian scenario presents that public universities receive mostly students who performed better and were prepared in elementary and high school education in private schools, while private universities attend students who received their basic education in public schools, which are characterized as low quality. These facts have led researchers to raise the possible determinants of student performance on standardized tests, such as the Brazilian Vestibular exam, to guide the development of policies aimed at equal access to College education. Seeking inspiration in North American models of affirmative action policies, some Brazilian public universities have suggested rate policies to enable and facilitate the entry of "minorities" (blacks, pardos1, natives, people of low income and public school students) to free College education. At the Federal University of the state Rio Grande do Norte (UFRN), the first incentives for candidates from public schools emerged in 2006, being improved and widespread during the last 7 years. This study aimed to analyse and discuss the Argument of Inclution (AI) - the affirmative action policy that provides additional scoring for students from public schools. From an extensive database, the Ordinary Least Squares (OLS) technique was used as well as a Quantile Regression considering as control the variables of personal, socioeconomic and educational characteristics of the candidates from the Brazilian Vestibular exam 2010 of the Federal University of the state Rio Grande do Norte (UFRN). The results demonstrate the importance of this incentive system, besides the magnitude of other variables

Endogamia em um rebanho da raça Guzerá

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Desempenho reprodutivo, características do pelame e taxa de sudação em vacas da raça Braford

Foram investigados os efeitos das características do pelame e da taxa de sudação (TS) sobre o intervalo de partos (IEP, n = 423) e os dias para o parto (DPP, n = 1.202) de vacas da raça Braford criadas em sistema extensivo em região tropical úmida no Mato Grosso do Sul, Brasil. As análises foram realizadas pelo método dos quadrados mínimos considerando para intervalo de partos os seguintes efeitos das classes de idade da vaca ao parto: grupo genético; pai; e regressão sobre a refletância da superfície da capa (R), a espessura da capa (E), o comprimento dos pêlos (C), o número de pêlos por unidade de área (N), o diâmetro dos pêlos (D) e a taxa de sudação. Para dias para o parto, foram considerados os efeitos: R, E, C, N, D, TS, divididos em classes; grupo genético; idade no início da estação de monta; e pai. Os componentes de variância e co-variância foram estimados pelo método de Máxima Verossimilhança Restrita sob o modelo touro. As características adaptativas não foram importantes para a variação do intervalo de parto e dos dias para o parto, mas a taxa de sudação apresentou efeito importante sobre dias ao parto. As estimativas de herdabilidade foram quase nulas para IEP e DPP e maiores para as características adaptativas (E = 0,16 ± 0,09; C = 0,18 ± 0,09; N = 0,08 ± 0,06; D = 0,12 ± 0,07; R = 0,30 ± 0,12; TS = 0,10 ± 0,06). Os resultados obtidos para as correlações genéticas de IEP e DPP com TS e entre as características do pelame indicaram que a seleção simultânea para menores valores de C pode ser favorável à redução do intervalo de partos. em geral, as correlações genéticas entre TS, IEP e DPP apresentaram elevado erro-padrão. Considerando os resultados deste estudo e a importância das características do pelame e da sudação, confirma-se a necessidade de se estudarem as correlações genéticas entre as características adaptativas e o desempenho reprodutivo para o progresso da adaptação de bovinos da raça Braford no Brasil.

Latent heat loss of dairy cows in an equatorial semi-arid environment

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)