A Hierarchical Approach for Multi-task Logistic Regression

Peer-reviewed

Proposição, validação e análise dos modelos que correlacionam estrutura química e atividade biológica

The present paper aims to bring under discussion some theoretical and practical aspects about the proposition, validation and analysis of QSAR models based on multiple linear regression. A comprehensive approach for the derivation of extrathermodynamic equations is reviewed. Some examples of QSAR models published in the literature are analyzed and criticized.

Learning Preferences with Kernel-Based Methods

Learning of preference relations has recently received significant attention in machine learning community. It is closely related to the classification and regression analysis and can be reduced to these tasks. However, preference learning involves prediction of ordering of the data points rather than prediction of a single numerical value as in case of regression or a class label as in case of classification. Therefore, studying preference relations within a separate framework facilitates not only better theoretical understanding of the problem, but also motivates development of the efficient algorithms for the task. Preference learning has many applications in domains such as information retrieval, bioinformatics, natural language processing, etc. For example, algorithms that learn to rank are frequently used in search engines for ordering documents retrieved by the query. Preference learning methods have been also applied to collaborative filtering problems for predicting individual customer choices from the vast amount of user generated feedback. In this thesis we propose several algorithms for learning preference relations. These algorithms stem from well founded and robust class of regularized least-squares methods and have many attractive computational properties. In order to improve the performance of our methods, we introduce several non-linear kernel functions. Thus, contribution of this thesis is twofold: kernel functions for structured data that are used to take advantage of various non-vectorial data representations and the preference learning algorithms that are suitable for different tasks, namely efficient learning of preference relations, learning with large amount of training data, and semi-supervised preference learning. Proposed kernel-based algorithms and kernels are applied to the parse ranking task in natural language processing, document ranking in information retrieval, and remote homology detection in bioinformatics domain. Training of kernel-based ranking algorithms can be infeasible when the size of the training set is large. This problem is addressed by proposing a preference learning algorithm whose computation complexity scales linearly with the number of training data points. We also introduce sparse approximation of the algorithm that can be efficiently trained with large amount of data. For situations when small amount of labeled data but a large amount of unlabeled data is available, we propose a co-regularized preference learning algorithm. To conclude, the methods presented in this thesis address not only the problem of the efficient training of the algorithms but also fast regularization parameter selection, multiple output prediction, and cross-validation. Furthermore, proposed algorithms lead to notably better performance in many preference learning tasks considered.

Analysis of laboratory intercomparison data: a matter of independence

When laboratory intercomparison exercises are conducted, there is no a priori dependence of the concentration of a certain compound determined in one laboratory to that determined by another(s). The same applies when comparing different methodologies. A existing data set of total mercury readings in fish muscle samples involved in a Brazilian intercomparison exercise was used to show that correlation analysis is the most effective statistical tool in this kind of experiments. Problems associated with alternative analytical tools such as mean or paired 't'-test comparison and regression analysis are discussed.

Financial stress transmission in EMU sovereign bond market volatility: a connectedness analysis

This paper measures the connectedness in EMU sovereign market volatility between April 1999 and January 2014, in order to monitor stress transmission and to identify episodes of intensive spillovers from one country to the others. To this end, we first perform a static and dynamic analysis to measure the total volatility connectedness in the entire period (the system-wide approach) using a framework recently proposed by Diebold and Yılmaz (2014). Second, we make use of a dynamic analysis to evaluate the net directional connectedness for each country and apply panel model techniques to investigate its determinants. Finally, to gain further insights, we examine the timevarying behaviour of net pair-wise directional connectedness at different stages of the recent sovereign debt crisis.

Association between Long-Term Exposure to Traffic-Related Air Pollution and Subclinical Atherosclerosis: The REGICOR Study

Background: Epidemiological evidence of the effects of long-term exposure to air pollu tion on the chronic processes of athero genesis is limited. Objective: We investigated the association of long-term exposure to traffic-related air pollu tion with subclinical atherosclerosis, measured by carotid intima media thickness (IMT) and ankle–brachial index (ABI). Methods: We performed a cross-sectional analysis using data collected during the reexamination (2007–2010) of 2,780 participants in the REGICOR (Registre Gironí del Cor: the Gerona Heart Register) study, a population-based prospective cohort in Girona, Spain. Long-term exposure across residences was calculated as the last 10 years’ time-weighted average of residential nitrogen dioxide (NO2) estimates (based on a local-scale land-use regression model), traffic intensity in the nearest street, and traffic intensity in a 100 m buffer. Associations with IMT and ABI were estimated using linear regression and multinomial logistic regression, respectively, controlling for sex, age, smoking status, education, marital status, and several other potential confounders or intermediates. Results: Exposure contrasts between the 5th and 95th percentiles for NO2 (25 μg/m), traffic intensity in the nearest street (15,000 vehicles/day), and traffic load within 100 m (7,200,000 vehicle-m/day) were associated with differences of 0.56% (95% CI: –1.5, 2.6%), 2.32% (95% CI: 0.48, 4.17%), and 1.91% (95% CI: –0.24, 4.06) percent difference in IMT, respectively. Exposures were positively associated with an ABI of > 1.3, but not an ABI of < 0.9. Stronger associations were observed among those with a high level of education and in men ≥ 60 years of age. Conclusions: Long-term traffic-related exposures were associated with subclinical markers of atherosclerosis. Prospective studies are needed to confirm associations and further examine differences among population subgroups.key words: ankle–brachial index, average daily traffic, cardiovascular disease, exposure assessment, exposure to tailpipe emissions, intima media thickness, land use regression model, Mediterranean diet, nitrogen dioxide

EU -jäsenyyden vaikutukset Latvian, Puolan, Unkarin ja Slovenian maataloussektoreiden tuottavuuskehitykseen

Tässä pro gradu -tutkielmassa tarkastellaan EU -jäsenyyden vaikutuksia itälaajentumisen myötä liittyneiden maiden maatalouteen ja sen tuottavuuteen. Maatalouden kehitys kuvaa kohdemaiden talouksien kehitystä. Uusien jäsenten kehitys taas vaikuttaa koko Euroopan unionin toimintaan ja sen asemaan maailmanmarkkinoilla. Tutkielman teoriaosuus esittelee tuottavuuden, yhteisen maatalouspolitiikan ja lineaarisen regressioanalyysin teoriaa. Empiriaosuudessa esitellään neljä kohdemaata ja tarkastellaan regressioanalyysien avulla sitä kuinka Euroopan unionin jäsenyys on vaikuttanut näiden maiden maataloussektoreiden tuottavuuteen.

On the Application of Beam on Elastic Foundation Theory to the Analysis of Stiffened Plate Strips

The main objective of this thesis is to show that plate strips subjected to transverse line loads can be analysed by using the beam on elastic foundation (BEF) approach. It is shown that the elastic behaviour of both the centre line section of a semi infinite plate supported along two edges, and the free edge of a cantilever plate strip can be accurately predicted by calculations based on the two parameter BEF theory. The transverse bending stiffness of the plate strip forms the foundation. The foundation modulus is shown, mathematically and physically, to be the zero order term of the fourth order differential equation governing the behaviour of BEF, whereas the torsion rigidity of the plate acts like pre tension in the second order term. Direct equivalence is obtained for harmonic line loading by comparing the differential equations of Levy's method (a simply supported plate) with the BEF method. By equating the second and zero order terms of the semi infinite BEF model for each harmonic component, two parameters are obtained for a simply supported plate of width B: the characteristic length, 1/ λ, and the normalized sum, n, being the effect of axial loading and stiffening resulting from the torsion stiffness, nlin. This procedure gives the following result for the first mode when a uniaxial stress field was assumed (ν = 0): 1/λ = √2B/π and nlin = 1. For constant line loading, which is the superimposition of harmonic components, slightly differing foundation parameters are obtained when the maximum deflection and bending moment values of the theoretical plate, with v = 0, and BEF analysis solutions are equated: 1 /λ= 1.47B/π and nlin. = 0.59 for a simply supported plate; and 1/λ = 0.99B/π and nlin = 0.25 for a fixed plate. The BEF parameters of the plate strip with a free edge are determined based solely on finite element analysis (FEA) results: 1/λ = 1.29B/π and nlin. = 0.65, where B is the double width of the cantilever plate strip. The stress biaxial, v > 0, is shown not to affect the values of the BEF parameters significantly the result of the geometric nonlinearity caused by in plane, axial and biaxial loading is studied theoretically by comparing the differential equations of Levy's method with the BEF approach. The BEF model is generalised to take into account the elastic rotation stiffness of the longitudinal edges. Finally, formulae are presented that take into account the effect of Poisson's ratio, and geometric non linearity, on bending behaviour resulting from axial and transverse inplane loading. It is also shown that the BEF parameters of the semi infinite model are valid for linear elastic analysis of a plate strip of finite length. The BEF model was verified by applying it to the analysis of bending stresses caused by misalignments in a laboratory test panel. In summary, it can be concluded that the advantages of the BEF theory are that it is a simple tool, and that it is accurate enough for specific stress analysis of semi infinite and finite plate bending problems.

Dinâmica não-linear e reagente limitante: um exemplo de bifurcação transcrítica

The nonlinear analysis of a general mixed second order reaction was performed, aiming to explore some basic tools concerning the mathematics of nonlinear differential equations. Concepts of stability around fixed points based on linear stability analysis are introduced, together with phase plane and integral curves. The main focus is the chemical relationship between changes of limiting reagent and transcritical bifurcation, and the investigation underlying the conclusion.

Critical-point yield model to estimate yield damage caused by Cercospora zea-maydis in corn

A model to estimate damage caused by gray leaf spot of corn (Cercospora zea-maydis) was developed from experimental field data gathered during the summer seasons of 2000/01 and during the second crop season [January-seedtime] of 2001, in the southwest of Goiás state. Three corn hybrids were grown over two seasons and on two sites, resulting in 12 experimental plots. A disease intensity gradient (lesions per leaf) was generated through application, three times over the season, of five different doses of the fungicide propiconazol. From tasseling onward, disease intensity on the ear leaf (El), and El - 1, El - 2, El + 1, and El + 2, was evaluated weekly. A manual harvest at the physiological ripening stage was followed by grain drying and cleaning. Finally, grain yield in kg.ha-1 was estimated. Regression analysis, performed between grain yield and all combinations of the number of lesions on each leaf type, generated thirty linear equations representing the damage function. To estimate losses caused by different disease intensities at different corn growth stages, these models should first be validated. Damage coefficients may be used in determining the economic damage threshold.

Analysis of Patterns in Electricity Spot Market Time Series

Due to its non-storability, electricity must be produced at the same time that it is consumed, as a result prices are determined on an hourly basis and thus analysis becomes more challenging. Moreover, the seasonal fluctuations in demand and supply lead to a seasonal behavior of electricity spot prices. The purpose of this thesis is to seek and remove all causal effects from electricity spot prices and remain with pure prices for modeling purposes. To achieve this we use Qlucore Omics Explorer (QOE) for the visualization and the exploration of the data set and Time Series Decomposition method to estimate and extract the deterministic components from the series. To obtain the target series we use regression based on the background variables (water reservoir and temperature). The result obtained is three price series (for Sweden, Norway and System prices) with no apparent pattern.

Estimation of standard enthalpy of formation of alkanes in gaseous state by calculating size, structural and electronic parameters in the molecules

A quantitative analysis is made on the correlation ship of thermodynamic property, i.e., standard enthalpy of formation (ΔH fº) with Kier's molecular connectivity index(¹Xv),vander waal's volume (Vw) electrotopological state index (E) and refractotopological state index (R) in gaseous state of alkanes. The regression analysis reveals a significant linear correlation of standard enthalpy of formation (ΔH fº) with ¹Xv, Vw, E and R. The equations obtained by regression analysis may be used to estimate standard enthalpy of formation (ΔH fº) of alkanes in gaseous state.

Critical-point model to estimate yield loss caused by Asian soybean rust

ABSTRACTA model to estimate yield loss caused by Asian soybean rust (ASR) (Phakopsora pachyrhizi) was developed by collecting data from field experiments during the growing seasons 2009/10 and 2010/11, in Passo Fundo, RS. The disease intensity gradient, evaluated in the phenological stages R5.3, R5.4 and R5.5 based on leaflet incidence (LI) and number of uredinium and lesions/cm2, was generated by applying azoxystrobin 60 g a.i/ha + cyproconazole 24 g a.i/ha + 0.5% of the adjuvant Nimbus. The first application occurred when LI = 25% and the remaining ones at 10, 15, 20 and 25-day intervals. Harvest occurred at physiological maturity and was followed by grain drying and cleaning. Regression analysis between the grain yield and the disease intensity assessment criteria generated 56 linear equations of the yield loss function. The greatest loss was observed in the earliest growth stage, and yield loss coefficients ranged from 3.41 to 9.02 kg/ha for each 1% LI for leaflet incidence, from 13.34 to 127.4 kg/ha/1 lesion/cm2 for lesion density and from 5.53 to 110.0 kg/ha/1 uredinium/cm2 for uredinium density.

Determination of the influence of the variation of reducing and non-reducing sugars on coffee quality with use of artificial neural network

The present study aimed at evaluating the use of Artificial Neural Network to correlate the values resulting from chemical analyses of samples of coffee with the values of their sensory analyses. The coffee samples used were from the Coffea arabica L., cultivars Acaiá do Cerrado, Topázio, Acaiá 474-19 and Bourbon, collected in the southern region of the state of Minas Gerais. The chemical analyses were carried out for reducing and non-reducing sugars. The quality of the beverage was evaluated by sensory analysis. The Artificial Neural Network method used values from chemical analyses as input variables and values from sensory analysis as output values. The multiple linear regression of sensory analysis values, according to the values from chemical analyses, presented a determination coefficient of 0.3106, while the Artificial Neural Network achieved a level of 80.00% of success in the classification of values from the sensory analysis.