Ratio selection for classification models

This paper is concerned with the selection of inputs for classification models based on ratios of measured quantities. For this purpose, all possible ratios are built from the quantities involved and variable selection techniques are used to choose a convenient subset of ratios. In this context, two selection techniques are proposed: one based on a pre-selection procedure and another based on a genetic algorithm. In an example involving the financial distress prediction of companies, the models obtained from ratios selected by the proposed techniques compare favorably to a model using ratios usually found in the financial distress literature.

On the utility of input selection and pruning for financial distress prediction models

Analyzes the use of linear and neural network models for financial distress classification, with emphasis on the issues of input variable selection and model pruning. A data-driven method for selecting input variables (financial ratios, in this case) is proposed. A case study involving 60 British firms in the period 1997-2000 is used for illustration. It is shown that the use of the Optimal Brain Damage pruning technique can considerably improve the generalization ability of a neural model. Moreover, the set of financial ratios obtained with the proposed selection procedure is shown to be an appropriate alternative to the ratios usually employed by practitioners.

A comparison of statistical methods for genomic selection in a mice population

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Classification and selection of biomarkers in genomic data using LASSO

High-throughput gene expression technologies such as microarrays have been utilized in a variety of scientific applications. Most of the work has been on assessing univariate associations between gene expression with clinical outcome (variable selection) or on developing classification procedures with gene expression data (supervised learning). We consider a hybrid variable selection/classification approach that is based on linear combinations of the gene expression profiles that maximize an accuracy measure summarized using the receiver operating characteristic curve. Under a specific probability model, this leads to consideration of linear discriminant functions. We incorporate an automated variable selection approach using LASSO. An equivalence between LASSO estimation with support vector machines allows for model fitting using standard software. We apply the proposed method to simulated data as well as data from a recently published prostate cancer study.

Time-Frequency based Feature Selection for Discrimination of non stationary Biosignals.

This research proposes a generic methodology for dimensionality reduction upon time-frequency representations applied to the classification of different types of biosignals. The methodology directly deals with the highly redundant and irrelevant data contained in these representations, combining a first stage of irrelevant data removal by variable selection, with a second stage of redundancy reduction using methods based on linear transformations. The study addresses two techniques that provided a similar performance: the first one is based on the selection of a set of the most relevant time?frequency points, whereas the second one selects the most relevant frequency bands. The first methodology needs a lower quantity of components, leading to a lower feature space; but the second improves the capture of the time-varying dynamics of the signal, and therefore provides a more stable performance. In order to evaluate the generalization capabilities of the methodology proposed it has been applied to two types of biosignals with different kinds of non-stationary behaviors: electroencephalographic and phonocardiographic biosignals. Even when these two databases contain samples with different degrees of complexity and a wide variety of characterizing patterns, the results demonstrate a good accuracy for the detection of pathologies, over 98%.The results open the possibility to extrapolate the methodology to the study of other biosignals.

Bayesian model selection of structural explanatory models: Application to road accident data

Using the Bayesian approach as the model selection criteria, the main purpose in this study is to establish a practical road accident model that can provide a better interpretation and prediction performance. For this purpose we are using a structural explanatory model with autoregressive error term. The model estimation is carried out through Bayesian inference and the best model is selected based on the goodness of fit measures. To cross validate the model estimation further prediction analysis were done. As the road safety measures the number of fatal accidents in Spain, during 2000-2011 were employed. The results of the variable selection process show that the factors explaining fatal road accidents are mainly exposure, economic factors, and surveillance and legislative measures. The model selection shows that the impact of economic factors on fatal accidents during the period under study has been higher compared to surveillance and legislative measures.

Distributed Feature Selection in Large n and Large p Regression Problems

Fitting statistical models is computationally challenging when the sample size or the dimension of the dataset is huge. An attractive approach for down-scaling the problem size is to first partition the dataset into subsets and then fit using distributed algorithms. The dataset can be partitioned either horizontally (in the sample space) or vertically (in the feature space), and the challenge arise in defining an algorithm with low communication, theoretical guarantees and excellent practical performance in general settings. For sample space partitioning, I propose a MEdian Selection Subset AGgregation Estimator ({\em message}) algorithm for solving these issues. The algorithm applies feature selection in parallel for each subset using regularized regression or Bayesian variable selection method, calculates the `median' feature inclusion index, estimates coefficients for the selected features in parallel for each subset, and then averages these estimates. The algorithm is simple, involves very minimal communication, scales efficiently in sample size, and has theoretical guarantees. I provide extensive experiments to show excellent performance in feature selection, estimation, prediction, and computation time relative to usual competitors.

While sample space partitioning is useful in handling datasets with large sample size, feature space partitioning is more effective when the data dimension is high. Existing methods for partitioning features, however, are either vulnerable to high correlations or inefficient in reducing the model dimension. In the thesis, I propose a new embarrassingly parallel framework named {\em DECO} for distributed variable selection and parameter estimation. In {\em DECO}, variables are first partitioned and allocated to m distributed workers. The decorrelated subset data within each worker are then fitted via any algorithm designed for high-dimensional problems. We show that by incorporating the decorrelation step, DECO can achieve consistent variable selection and parameter estimation on each subset with (almost) no assumptions. In addition, the convergence rate is nearly minimax optimal for both sparse and weakly sparse models and does NOT depend on the partition number m. Extensive numerical experiments are provided to illustrate the performance of the new framework.

For datasets with both large sample sizes and high dimensionality, I propose a new "divided-and-conquer" framework {\em DEME} (DECO-message) by leveraging both the {\em DECO} and the {\em message} algorithm. The new framework first partitions the dataset in the sample space into row cubes using {\em message} and then partition the feature space of the cubes using {\em DECO}. This procedure is equivalent to partitioning the original data matrix into multiple small blocks, each with a feasible size that can be stored and fitted in a computer in parallel. The results are then synthezied via the {\em DECO} and {\em message} algorithm in a reverse order to produce the final output. The whole framework is extremely scalable.

Práticas de alimentação complementar no primeiro ano de vida e fatores associados

OBJETIVO: Investigar a prevalência de consumo de alimentos complementares e os fatores associados à alimentação complementar oportuna em menores de um ano. MÉTODOS: Participaram do estudo 1 176 crianças, durante a Campanha Nacional de Vacinação de 2003, em São Bernardo do Campo (SP), cujos acompanhantes responderam questionário que incluiu questões sobre a alimentação da criança nas 24 horas precedentes. A estimativa da prevalência de consumo dos alimentos complementares foi realizada por um modelo de regressão logística ajustado por idade; as medianas de introdução de alimentos por análise de sobrevida e os fatores associados à alimentação complementar oportuna por regressão de Poisson com ajuste robusto de variância e seleção hierarquizada de variáveis. RESULTADOS: Observou-se introdução precoce de alimentos complementares: no quarto mês, cerca de um terço das crianças recebiam suco de fruta e um quarto das crianças recebiam mingau, fruta ou sopa, ao passo que a probabilidade de consumir a comida da família aos oito meses foi baixa (48%). A mediana de idade para o consumo de frutas foi de 266 dias (IC95% 256-275), de papa de legumes foi 258 dias (IC95% 250-264) e comida da família, 292 dias (IC 95% 287-303). Os fatores associados ao consumo de alimentos sólidos antes dos seis meses de idade foram: sistema de assistência à saúde; idade materna; trabalho materno e uso de chupeta. CONCLUSÃO: O consumo precoce de alimentos sólidos, um risco potencial para a saúde infantil e para o desenvolvimento de doenças crônicas na idade adulta, evidenciam a necessidade de ações programáticas para reversão deste quadro.

Comparison of univariate and multivariate calibration for the determination of micronutrients in pellets of plant materials by laser induced breakdown spectrometry

The application of laser induced breakdown spectrometry (LIBS) aiming the direct analysis of plant materials is a great challenge that still needs efforts for its development and validation. In this way, a series of experimental approaches has been carried out in order to show that LIBS can be used as an alternative method to wet acid digestions based methods for analysis of agricultural and environmental samples. The large amount of information provided by LIBS spectra for these complex samples increases the difficulties for selecting the most appropriated wavelengths for each analyte. Some applications have suggested that improvements in both accuracy and precision can be achieved by the application of multivariate calibration in LIBS data when compared to the univariate regression developed with line emission intensities. In the present work, the performance of univariate and multivariate calibration, based on partial least squares regression (PLSR), was compared for analysis of pellets of plant materials made from an appropriate mixture of cryogenically ground samples with cellulose as the binding agent. The development of a specific PLSR model for each analyte and the selection of spectral regions containing only lines of the analyte of interest were the best conditions for the analysis. In this particular application, these models showed a similar performance. but PLSR seemed to be more robust due to a lower occurrence of outliers in comparison to the univariate method. Data suggests that efforts dealing with sample presentation and fitness of standards for LIBS analysis must be done in order to fulfill the boundary conditions for matrix independent development and validation. (C) 2009 Elsevier B.V. All rights reserved.

Quantification of chlorpheniramine maleate enantiomers by ultraviolet spectroscopy and chemometric methods

Chlorpheniramine maleate (CLOR) enantiomers were quantified by ultraviolet spectroscopy and partial least squares regression. The CLOR enantiomers were prepared as inclusion complexes with beta-cyclodextrin and 1-butanol with mole fractions in the range from 50 to 100%. For the multivariate calibration the outliers were detected and excluded and variable selection was performed by interval partial least squares and a genetic algorithm. Figures of merit showed results for accuracy of 3.63 and 2.83% (S)-CLOR for root mean square errors of calibration and prediction, respectively. The ellipse confidence region included the point for the intercept and the slope of 1 and 0, respectively. Precision and analytical sensitivity were 0.57 and 0.50% (S)-CLOR, respectively. The sensitivity, selectivity, adjustment, and signal-to-noise ratio were also determined. The model was validated by a paired t test with the results obtained by high-performance liquid chromatography proposed by the European pharmacopoeia and circular dichroism spectroscopy. The results showed there was no significant difference between the methods at the 95% confidence level, indicating that the proposed method can be used as an alternative to standard procedures for chiral analysis.

QSAR modeling of antitubercular activity of diverse organic compounds

Tuberculosis (TB) is a worldwide infectious disease that has shown over time extremely high mortality levels. The urgent need to develop new antitubercular drugs is due to the increasing rate of appearance of multi-drug resistant strains to the commonly used drugs, and the longer durations of therapy and recovery, particularly in immuno-compromised patients. The major goal of the present study is the exploration of data from different families of compounds through the use of a variety of machine learning techniques so that robust QSAR-based models can be developed to further guide in the quest for new potent anti-TB compounds. Eight QSAR models were built using various types of descriptors (from ADRIANA.Code and Dragon software) with two publicly available structurally diverse data sets, including recent data deposited in PubChem. QSAR methodologies used Random Forests and Associative Neural Networks. Predictions for the external evaluation sets obtained accuracies in the range of 0.76-0.88 (for active/inactive classifications) and Q(2)=0.66-0.89 for regressions. Models developed in this study can be used to estimate the anti-TB activity of drug candidates at early stages of drug development (C) 2011 Elsevier B.V. All rights reserved.

Azorean Agriculture Efficiency by PAR

Copyright © 2013 Springer Netherlands.

Cluster analysis and segmentation of Global M&A transactions

A Work Project, presented as part of the requirements for the Award of a Masters Degree in Finance from the NOVA – School of Business and Economics

Feature Nets: behavioural modelling of software product lines

Software product lines (SPL) are diverse systems that are developed using a dual engineering process: (a)family engineering defines the commonality and variability among all members of the SPL, and (b) application engineering derives specific products based on the common foundation combined with a variable selection of features. The number of derivable products in an SPL can thus be exponential in the number of features. This inherent complexity poses two main challenges when it comes to modelling: Firstly, the formalism used for modelling SPLs needs to be modular and scalable. Secondly, it should ensure that all products behave correctly by providing the ability to analyse and verify complex models efficiently. In this paper we propose to integrate an established modelling formalism (Petri nets) with the domain of software product line engineering. To this end we extend Petri nets to Feature Nets. While Petri nets provide a framework for formally modelling and verifying single software systems, Feature Nets offer the same sort of benefits for software product lines. We show how SPLs can be modelled in an incremental, modular fashion using Feature Nets, provide a Feature Nets variant that supports modelling dynamic SPLs, and propose an analysis method for SPL modelled as Feature Nets. By facilitating the construction of a single model that includes the various behaviours exhibited by the products in an SPL, we make a significant step towards efficient and practical quality assurance methods for software product lines.

Genome-wide patterns of latitudinal differentiation among populations of Drosophila melanogaster from North America.

Understanding the genetic underpinnings of adaptive change is a fundamental but largely unresolved problem in evolutionary biology. Drosophila melanogaster, an ancestrally tropical insect that has spread to temperate regions and become cosmopolitan, offers a powerful opportunity for identifying the molecular polymorphisms underlying clinal adaptation. Here, we use genome-wide next-generation sequencing of DNA pools ('pool-seq') from three populations collected along the North American east coast to examine patterns of latitudinal differentiation. Comparing the genomes of these populations is particularly interesting since they exhibit clinal variation in a number of important life history traits. We find extensive latitudinal differentiation, with many of the most strongly differentiated genes involved in major functional pathways such as the insulin/TOR, ecdysone, torso, EGFR, TGFβ/BMP, JAK/STAT, immunity and circadian rhythm pathways. We observe particularly strong differentiation on chromosome 3R, especially within the cosmopolitan inversion In(3R)Payne, which contains a large number of clinally varying genes. While much of the differentiation might be driven by clinal differences in the frequency of In(3R)P, we also identify genes that are likely independent of this inversion. Our results provide genome-wide evidence consistent with pervasive spatially variable selection acting on numerous loci and pathways along the well-known North American cline, with many candidates implicated in life history regulation and exhibiting parallel differentiation along the previously investigated Australian cline.