On cluster analysis of complex and heterogeneous data

3rd SMTDA Conference Proceedings, 11-14 June 2014, Lisbon Portugal.

Sistema de recomendação para plataformas de e-learning

O aumento do número de recursos digitais disponíveis dificulta a tarefa de pesquisa dos recursos mais relevantes, no sentido de se obter o que é mais relevante. Assim sendo, um novo tipo de ferramentas, capaz de recomendar os recursos mais apropriados às necessidades do utilizador, torna-se cada vez mais necessário. O objetivo deste trabalho de I&D é o de implementar um módulo de recomendação inteligente para plataformas de e-learning. As recomendações baseiam-se, por um lado, no perfil do utilizador durante o processo de formação e, por outro lado, nos pedidos efetuados pelo utilizador, através de pesquisas [Tavares, Faria e Martins, 2012]. O e-learning 3.0 é um projeto QREN desenvolvido por um conjunto de organizações e tem com objetivo principal implementar uma plataforma de e-learning. Este trabalho encontra-se inserido no projeto e-learning 3.0 e consiste no desenvolvimento de um módulo de recomendação inteligente (MRI). O MRI utiliza diferentes técnicas de recomendação já aplicadas noutros sistemas de recomendação. Estas técnicas são utilizadas para criar um sistema de recomendação híbrido direcionado para a plataforma de e-learning. Para representar a informação relevante, sobre cada utilizador, foi construído um modelo de utilizador. Toda a informação necessária para efetuar a recomendação será representada no modelo do utilizador, sendo este modelo atualizado sempre que necessário. Os dados existentes no modelo de utilizador serão utilizados para personalizar as recomendações produzidas. As recomendações estão divididas em dois tipos, a formal e a não formal. Na recomendação formal o objetivo é fazer sugestões relacionadas a um curso específico. Na recomendação não-formal, o objetivo é fazer sugestões mais abrangentes onde as recomendações não estão associadas a nenhum curso. O sistema proposto é capaz de sugerir recursos de aprendizagem, com base no perfil do utilizador, através da combinação de técnicas de similaridade de palavras, um algoritmo de clustering e técnicas de filtragem [Tavares, Faria e Martins, 2012].

Os efeitos das variações das taxas de juro na volatilidade dos mercados bolsistas

Mestrado em Controlo de Gestão e dos Negócios

Entrepreneurship Education : The role of the Higher Education Institutions in the Entrepreneurial Attitudes of the Students

8th International Conference of Education, Research and Innovation. 18-20 November, 2015, Seville, Spain.

Agrupamento de dados visual interactivo

Com a crescente geração, armazenamento e disseminação da informação nos últimos anos, o anterior problema de falta de informação transformou-se num problema de extracção do conhecimento útil a partir da informação disponível. As representações visuais da informação abstracta têm sido utilizadas para auxiliar a interpretação os dados e para revelar padrões de outra forma escondidos. A visualização de informação procura aumentar a cognição humana aproveitando as capacidades visuais humanas, de forma a tornar perceptível a informação abstracta, fornecendo os meios necessários para que um humano possa absorver quantidades crescentes de informação, com as suas capacidades de percepção. O objectivo das técnicas de agrupamento de dados consiste na divisão de um conjunto de dados em vários grupos, em que dados semelhantes são colocados no mesmo grupo e dados dissemelhantes em grupos diferentes. Mais especificamente, o agrupamento de dados com restrições tem o intuito de incorporar conhecimento a priori no processo de agrupamento de dados, com o objectivo de aumentar a qualidade do agrupamento de dados e, simultaneamente, encontrar soluções apropriadas a tarefas e interesses específicos. Nesta dissertação é estudado a abordagem de Agrupamento de Dados Visual Interactivo que permite ao utilizador, através da interacção com uma representação visual da informação, incorporar o seu conhecimento prévio acerca do domínio de dados, de forma a influenciar o agrupamento resultante para satisfazer os seus objectivos. Esta abordagem combina e estende técnicas de visualização interactiva de informação, desenho de grafos de forças direccionadas e agrupamento de dados com restrições. Com o propósito de avaliar o desempenho de diferentes estratégias de interacção com o utilizador, são efectuados estudos comparativos utilizando conjuntos de dados sintéticos e reais.

Surveillance of mother-to-child HIV transmission: socioeconomic and health care coverage indicators

OBJECTIVE: To identify clustering areas of infants exposed to HIV during pregnancy and their association with indicators of primary care coverage and socioeconomic condition. METHODS: Ecological study where the unit of analysis was primary care coverage areas in the city of Porto Alegre, Southern Brazil, in 2003. Geographical Information System and spatial analysis tools were used to describe indicators of primary care coverage areas and socioeconomic condition, and estimate the prevalence of liveborn infants exposed to HIV during pregnancy and delivery. Data was obtained from Brazilian national databases. The association between different indicators was assessed using Spearman's nonparametric test. RESULTS: There was found an association between HIV infection and high birth rates (r=0.22, p<0.01) and lack of prenatal care (r=0.15, p<0.05). The highest HIV infection rates were seen in areas with poor socioeconomic conditions and difficult access to health services (r=0.28, p<0.01). The association found between higher rate of prenatal care among HIV-infected women and adequate immunization coverage (r=0.35, p<0.01) indicates that early detection of HIV infection is effective in those areas with better primary care services. CONCLUSIONS: Urban poverty is a strong determinant of mother-to-child HIV transmission but this trend can be fought with health surveillance at the primary care level.

Dominant set approach to ECG biometrics

Electrocardiographic (ECG) signals are emerging as a recent trend in the field of biometrics. In this paper, we propose a novel ECG biometric system that combines clustering and classification methodologies. Our approach is based on dominant-set clustering, and provides a framework for outlier removal and template selection. It enhances the typical workflows, by making them better suited to new ECG acquisition paradigms that use fingers or hand palms, which lead to signals with lower signal to noise ratio, and more prone to noise artifacts. Preliminary results show the potential of the approach, helping to further validate the highly usable setups and ECG signals as a complementary biometric modality.

Analysis and visualization of seismic data using mutual information

Seismic data is difficult to analyze and classical mathematical tools reveal strong limitations in exposing hidden relationships between earthquakes. In this paper, we study earthquake phenomena in the perspective of complex systems. Global seismic data, covering the period from 1962 up to 2011 is analyzed. The events, characterized by their magnitude, geographic location and time of occurrence, are divided into groups, either according to the Flinn-Engdahl (F-E) seismic regions of Earth or using a rectangular grid based in latitude and longitude coordinates. Two methods of analysis are considered and compared in this study. In a first method, the distributions of magnitudes are approximated by Gutenberg-Richter (G-R) distributions and the parameters used to reveal the relationships among regions. In the second method, the mutual information is calculated and adopted as a measure of similarity between regions. In both cases, using clustering analysis, visualization maps are generated, providing an intuitive and useful representation of the complex relationships that are present among seismic data. Such relationships might not be perceived on classical geographic maps. Therefore, the generated charts are a valid alternative to other visualization tools, for understanding the global behavior of earthquakes.

Fractional dynamics and MDS visualization of earthquake phenomena

This paper analyses earthquake data in the perspective of dynamical systems and fractional calculus (FC). This new standpoint uses Multidimensional Scaling (MDS) as a powerful clustering and visualization tool. FC extends the concepts of integrals and derivatives to non-integer and complex orders. MDS is a technique that produces spatial or geometric representations of complex objects, such that those objects that are perceived to be similar in some sense are placed on the MDS maps forming clusters. In this study, over three million seismic occurrences, covering the period from January 1, 1904 up to March 14, 2012 are analysed. The events are characterized by their magnitude and spatiotemporal distributions and are divided into fifty groups, according to the Flinn–Engdahl (F–E) seismic regions of Earth. Several correlation indices are proposed to quantify the similarities among regions. MDS maps are proven as an intuitive and useful visual representation of the complex relationships that are present among seismic events, which may not be perceived on traditional geographic maps. Therefore, MDS constitutes a valid alternative to classic visualization tools for understanding the global behaviour of earthquakes.

Determining the number of clusters in categorical data

Cluster analysis for categorical data has been an active area of research. A well-known problem in this area is the determination of the number of clusters, which is unknown and must be inferred from the data. In order to estimate the number of clusters, one often resorts to information criteria, such as BIC (Bayesian information criterion), MML (minimum message length, proposed by Wallace and Boulton, 1968), and ICL (integrated classification likelihood). In this work, we adopt the approach developed by Figueiredo and Jain (2002) for clustering continuous data. They use an MML criterion to select the number of clusters and a variant of the EM algorithm to estimate the model parameters. This EM variant seamlessly integrates model estimation and selection in a single algorithm. For clustering categorical data, we assume a finite mixture of multinomial distributions and implement a new EM algorithm, following a previous version (Silvestre et al., 2008). Results obtained with synthetic datasets are encouraging. The main advantage of the proposed approach, when compared to the above referred criteria, is the speed of execution, which is especially relevant when dealing with large data sets.

Cardiorespiratory fitness is negatively associated with metabolic risk factors independently of the adherence to a healthy dietary pattern

Background and aim: Cardiorespiratory fitness (CRF) and diet have been involved as significant factors towards the prevention of cardio-metabolic diseases. This study aimed to assess the impact of the combined associations of CRF and adherence to the Southern European Atlantic Diet (SEADiet) on the clustering of metabolic risk factors in adolescents. Methods and Results: A cross-sectional school-based study was conducted on 468 adolescents aged 15-18, from the Azorean Islands, Portugal. We measured fasting glucose, insulin, total cholesterol (TC), HDL-cholesterol, triglycerides, systolic blood pressure, waits circumference and height. HOMA, TC/HDL-C ratio and waist-to-height ratio were calculated. For each of these variables, a Z-score was computed by age and sex. A metabolic risk score (MRS) was constructed by summing the Z scores of all individual risk factors. High risk was considered when the individual had 1SD of this score. CRF was measured with the 20 m-Shuttle-Run- Test. Adherence to SEADiet was assessed with a semi-quantitative food frequency questionnaire. Logistic regression showed that, after adjusting for potential confounders, unfit adolescents with low adherence to SEADiet had the highest odds of having MRS (OR Z 9.4; 95%CI:2.6e33.3) followed by the unfit ones with high adherence to the SEADiet (OR Z 6.6; 95% CI: 1.9e22.5) when compared to those who were fit and had higher adherence to SEADiet.

Cluster de volatilidade no mercado de taxas de câmbio USD/GBP

Mestrado em Controlo de Gestão e dos Negócios

Optimal virtual cluster-based multiprocessor scheduling

Scheduling of constrained deadline sporadic task systems on multiprocessor platforms is an area which has received much attention in the recent past. It is widely believed that finding an optimal scheduler is hard, and therefore most studies have focused on developing algorithms with good processor utilization bounds. These algorithms can be broadly classified into two categories: partitioned scheduling in which tasks are statically assigned to individual processors, and global scheduling in which each task is allowed to execute on any processor in the platform. In this paper we consider a third, more general, approach called cluster-based scheduling. In this approach each task is statically assigned to a processor cluster, tasks in each cluster are globally scheduled among themselves, and clusters in turn are scheduled on the multiprocessor platform. We develop techniques to support such cluster-based scheduling algorithms, and also consider properties that minimize total processor utilization of individual clusters. In the last part of this paper, we develop new virtual cluster-based scheduling algorithms. For implicit deadline sporadic task systems, we develop an optimal scheduling algorithm that is neither Pfair nor ERfair. We also show that the processor utilization bound of us-edf{m/(2m−1)} can be improved by using virtual clustering. Since neither partitioned nor global strategies dominate over the other, cluster-based scheduling is a natural direction for research towards achieving improved processor utilization bounds.

Prevalence and determinants of child undernutrition and stunting in semiarid region of Brazil

OBJECTIVE : To analyze the evolution in the prevalence and determinants of malnutrition in children in the semiarid region of Brazil. METHODS : Data were collected from two cross-sectional population-based household surveys that used the same methodology. Clustering sampling was used to collect data from 8,000 families in Ceará, Northeastern Brazil, for the years 1987 and 2007. Acute undernutrition was calculated as weight/age < -2 standard deviation (SD); stunting as height/age < -2 SD; wasting as weight/height < -2 SD. Data on biological and sociodemographic determinants were analyzed using hierarchical multivariate analyses based on a theoretical model. RESULTS : A sample of 4,513 and 1,533 children under three years of age, in 1987 and 2007, respectively, were included in the analyses. The prevalence of acute malnutrition was reduced by 60.0%, from 12.6% in 1987 to 4.7% in 2007, while prevalence of stunting was reduced by 50.0%, from 27.0% in 1987 to 13.0% in 2007. Prevalence of wasting changed little in the period. In 1987, socioeconomic and biological characteristics (family income, mother’s education, toilet and tap water availability, children’s medical consultation and hospitalization, age, sex and birth weight) were significantly associated with undernutrition, stunting and wasting. In 2007, the determinants of malnutrition were restricted to biological characteristics (age, sex and birth weight). Only one socioeconomic characteristic, toilet availability, remained associated with stunting. CONCLUSIONS : Socioeconomic development, along with health interventions, may have contributed to improvements in children’s nutritional status. Birth weight, especially extremely low weight (< 1,500 g), appears as the most important risk factor for early childhood malnutrition.

ECG Biometrics: A Template Selection Approach

Electrocardiography (ECG) biometrics is emerging as a viable biometric trait. Recent developments at the sensor level have shown the feasibility of performing signal acquisition at the fingers and hand palms, using one-lead sensor technology and dry electrodes. These new locations lead to ECG signals with lower signal to noise ratio and more prone to noise artifacts; the heart rate variability is another of the major challenges of this biometric trait. In this paper we propose a novel approach to ECG biometrics, with the purpose of reducing the computational complexity and increasing the robustness of the recognition process enabling the fusion of information across sessions. Our approach is based on clustering, grouping individual heartbeats based on their morphology. We study several methods to perform automatic template selection and account for variations observed in a person's biometric data. This approach allows the identification of different template groupings, taking into account the heart rate variability, and the removal of outliers due to noise artifacts. Experimental evaluation on real world data demonstrates the advantages of our approach.