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.

Visualization of clusters in geo-referenced data using three-dimensional self-organizing maps

Dissertação apresentada como requisito parcial para obtenção do grau de Mestre em Estatística e Gestão de Informação

PERFINOV: Perfis para a inovação: uma abordagem baseada nos clusters

Desenvolvido no âmbito do Programa PO Norte, o presente estudo, designado de PERFINOV – PERFIS PARA A INOVAÇÃO – Uma Abordagem Baseada nos Clusters apresenta, como principal objectivo proceder à “Definição de modelos e metodologias de abordagens baseadas em clusters, decorrentes dos princípios das políticas de apoio à inovação, visando a identificação de necessi- dades ao nível das competências e qualificações actuais e futuras das empresas neles inseridos, apelando às relações sinérgicas fruto do relacionamento, tendo em vista o incremento do valor acrescentado dos mesmos, a sua ascensão na cadeia de valor e a promoção do bottom-up do de- senvolvimento regional.” (UERN, Memória descritiva, 2007)

Um middleware para computação paralela em clusters de multicores

Dissertação para obtenção do Grau de Mestre em Engenharia Informática

Ligand K-edge X-ray absorption spectroscopy and DFT calculations on [Fe3S4]0,+ clusters: delocalization, redox, and effect of the protein environment

Ligand K-edge XAS of an [Fe3S4]0 model complex is reported. The pre-edge can be resolved into contributions from the í2Ssulfide, í3Ssulfide, and Sthiolate ligands. The average ligand-metal bond covalencies obtained from these pre-edges are further distributed between Fe3+ and Fe2.5+ components using DFT calculations. The bridging ligand covalency in the [Fe2S2]+ subsite of the [Fe3S4]0 cluster is found to be significantly lower than its value in a reduced [Fe2S2] cluster (38% vs 61%, respectively). This lowered bridging ligand covalency reduces the superexchange coupling parameter J relative to its value in a reduced [Fe2S2]+ site (-146 cm-1 vs -360 cm-1, respectively). This decrease in J, along with estimates of the double exchange parameter B and vibronic coupling parameter ì2/k-, leads to an S ) 2 delocalized ground state in the [Fe3S4]0 cluster. The S K-edge XAS of the protein ferredoxin II (Fd II) from the D. gigas active site shows a decrease in covalency compared to the model complex, in the same oxidation state, which correlates with the number of H-bonding interactions to specific sulfur ligands present in the active site. The changes in ligand-metal bond covalencies upon redox compared with DFT calculations indicate that the redox reaction involves a two-electron change (one-electron ionization plus a spin change of a second electron) with significant electronic relaxation. The presence of the redox inactive Fe3+ center is found to decrease the barrier of the redox process in the [Fe3S4] cluster due to its strong antiferromagnetic coupling with the redox active Fe2S2 subsite.

Building clusters for CRM strategies by mining airlines customer data

Trabalho de Projeto apresentado como requisito parcial para obtenção do grau de Mestre em Estatística e Gestão de Informação

Criação de clusters de clientes de um banco através do self-organizing maps hierárquico: comparação entre HSOM e SOM

Trabalho de Projeto apresentado como requisito parcial para obtenção do grau de Mestre em Estatística e Gestão de Informação

How can clusters help in the product type choice? The case of a specialized retail store

This work project (WP) is a study about a clustering strategy for Sport Zone. The general cluster study’s objective is to create groups such that within each group the individuals are similar to each other, but should be different among groups. The clusters creation is a mix of common sense, trial and error and some statistical supporting techniques. Our particular objective is to support category managers to better define the product type to be displayed in the stores’ shelves by doing store clusters. This research was carried out for Sport Zone, and comprises an objective definition, a literature review, the clustering activity itself, some factor analysis and a discriminant analysis to better frame our work. Together with this quantitative part, a survey addressed to category managers to better understand their key drivers, for choosing the type of product of each store, was carried out. Based in a non-random sample of 65 stores with data referring to 2013, the final result was the choice of 6 store clusters (Figure 1) which were individually characterized as the main outcome of this work. In what relates to our selected variables, all were important for the distinction between clusters, which proves the adequacy of their choice. The interpretation of the results gives category managers a tool to understand which products best fit the clustered stores. Furthermore, as a side finding thanks to the clusterization, a STP (Segmentation, Targeting and Positioning) was initiated, being this WP the first steps of a continuous process.

Pulmonary tuberculosis in São Luis, State of Maranhão, Brazil: space and space-time risk clusters for death (2008-2012)

INTRODUCTION: The objective was to identify space and space-time risk clusters for the occurrence of deaths in a priority city for the control of tuberculosis (TB) in the Brazilian Northeast. METHODS: Ecological research was undertaken in the City of São Luis/Maranhão. Cases were considered that resulted in deaths in the population living in the urban region of the city with pulmonary TB as the basic cause, between 2008 and 2012. To detect space and space-time clusters of deaths due to pulmonary TB in the census sectors, the spatial analysis scan technique was used. RESULTS: In total, 221 deaths by TB occurred, 193 of which were due to pulmonary TB. Approximately 95% of the cases (n=183) were geocoded. Two significant spatial clusters were identified, the first of which showed a mortality rate of 5.8 deaths per 100,000 inhabitants per year and a high relative risk of 3.87. The second spatial cluster showed a mortality rate of 0.4 deaths per 100,000 inhabitants per year and a low relative risk of 0.10. A significant cluster was observed in the space-time analysis between 11/01/2008 and 04/30/2011, with a mortality rate of 8.10 deaths per 100,000 inhabitants per year and a high relative risk (3.0). CONCLUSIONS: The knowledge of priority sites for the occurrence of deaths can support public management to reduce inequities in the access to health services and permit an optimization of the resources and teams in the control of pulmonary TB, providing support for specific strategies focused on the most vulnerable populations.

Resonancia Magnética Nuclear aplicada al estudio de nuevos sistemas farmacéuticos y biológicos, complejos poliméricos y conjuntos de espines modelo. Nuclear Magnetic Resonance applied to the study of new pharmaceutical and biological systems, polymeric complexes and model spin clusters.

El presente proyecto plantea utilizar integralmente la técnica de Resonancia Magnética Nuclear en sólidos como un medio experimental que permite entender fenómenos de la física fundamental, como así también realizar aplicaciones de interés en el campo de la química, los desarrollos farmacéuticos y la biología. Novedosas técnicas experimentales serán empleadas, en conjunción con otras más tradicionales, en la caracterización de nuevas estructuras poliméricas acomplejadas a metales, membranas biológicas y compuestos de interés farmacéutico en vías de desarrollo, los cuales presentan el fenómeno de polimorfismo . Esto se llevará a cabo complementando los resultados de RMN en sólidos con técnicas tanto espectroscópicas como analíticas (Infrarrojo, Difracción de Rayos X, Calorimetría, RMN en solución) y trabajo interdisciplinario. Paralelamente al desarrollo de estos temas, profundizaremos mediante investigación básica, en la compresión de la dinámica cuántica y el surgimiento de la irreversibilidad en sistemas de espines nucleares. Observaremos en particular la generación, evolución y control de las coherencias cuánticas múltiples en sistemas cuánticos abiertos, lo cual nos da información sobre tamaño de clusters de espines. Esto permitirá la correcta implementación de secuencias de pulsos sofisticadas, como así también desarrollar nuevos métodos de medición aplicados a la caracterización estructural y a la dinámica molecular de sólidos complejos. Debemos resaltar que este proyecto está conectado con aspectos tanto básicos como aplicados de la RMN en sólidos como técnica experimental, la cual se desarrolla en el país únicamente en FaMAF-UNC. Se nutre además de trabajo multidisciplinario promoviendo la colaboración con investigadores y becarios de distintas áreas (física, química, farmacia, biología) provenientes de distintos puntos del país. Finalmente podemos afirmar que este plan impulsa la aplicación de la física básica proyectada a diferentes áreas del conocimiento, en el ámbito de la provincia de Córdoba. The aim of the present proyect is to use Nuclear Magnetic Resonance (NMR) as a complete techique that allows the understanding of fundamental physics phenomena and, at the same time, it leads to important applications in the fields of chemistry, pharmaceutical developments and biology. New experiments will be used together with traditional ones, in the characterization of new metal-polymer complexes, biological membranes and pharmaceutical compounds, some of them presenting polymorfism. NMR experiments will be complemented with diverse spectroscopic and analytical techniques: Infrared, X ray Diffraction, Thermal Analysis, solution NMR, as well as multidisciplinary investigation. Additionally, the present proyect plans to study in depth several aspects of quantum dynamics phenomena and decoherence in nuclear spin systems. The present proyect is connected with basic and applied aspects of the solid state NMR technique, developed in our country, only at FaMAF-UNC. It is is composed by multidisciplinary work and it promotes the collaboration with researchers and students coming from different fields (physics, chemistry, pharmaceutical developments, biology) and different points of our country.

Entorno de desarrollo para clusters

El presente trabajo expone los entornos de desarrollo de aplicaciones paralelas en sistemas distribuidos. Se ha dedicado especial atención a los entornos gráficos, como plataforma de acceso al cluster. Dada la cantidad de alternativas a seleccionar en este tipo de entornos, como puede ser la parametrización de las aplicaciones, especificación de la arquitectura a utilizar de los nodos, caracterización de la carga, políticas de planificación, etc, hace que este trabajo pueda servir de punto inicial de partida a usuarios que deseen iniciarse en este tipo de entornos.

Firms in Scottish High Technology Clusters: software, life sciences, microelectronics, optoelectronics and digital media – preliminary evidence and analysis on firm size, growth and optimality

This paper reports on: (a) new primary source evidence on; and (b) statistical and econometric analysis of high technology clusters in Scotland. It focuses on the following sectors: software, life sciences, microelectronics, optoelectronics, and digital media. Evidence on a postal and e-mailed questionnaire is presented and discussed under the headings of: performance, resources, collaboration & cooperation, embeddedness, and innovation. The sampled firms are characterised as being small (viz. micro-firms and SMEs), knowledge intensive (largely graduate staff), research intensive (mean spend on R&D GBP 842k), and internationalised (mainly selling to markets beyond Europe). Preliminary statistical evidence is presented on Gibrat’s Law (independence of growth and size) and the Schumpeterian Hypothesis (scale economies in R&D). Estimates suggest a short-run equilibrium size of just 100 employees, but a long-run equilibrium size of 1000 employees. Further, to achieve the Schumpeterian effect (of marked scale economies in R&D), estimates suggest that firms have to grow to very much larger sizes of beyond 3,000 employees. We argue that the principal way of achieving the latter scale may need to be by takeovers and mergers, rather than by internally driven growth.

Manejo de históricos en clusters

CISNE es un sistema de cómputo en paralelo del Departamento de Arquitectura de Computadores y Sistemas Operativos (DACSO). Para poder implementar políticas de ordenacción de colas y selección de trabajos, este sistema necesita predecir el tiempo de ejecución de las aplicaciones. Con este trabajo se pretende proveer al sistema CISNE de un método para predecir el tiempo de ejecución basado en un histórico donde se almacenarán todos los datos sobre las ejecuciones.

The Significance of Identifying Industrial clusters The Case of Scotland

Industrial clustering policy is now an integral part of economic development planning in most advanced economies. However, there have been concerns in some quarters over the ability of an industrial cluster-based development strategy to deliver its promised economic benefits and this has been increasingly been blamed on the failure by governments to identify industrial clusters. In a study published in 2001, the DTI identified clusters across the UK based on the comparative scale and significance of industrial sectors. The study identified thirteen industrial clusters in Scotland. However the clusters identified are not a homogeneous set and they seem to vary in terms of their geographic concentration within Scotland. This paper examines the spatial distribution of industries within Scotland, thereby identifying more localised clusters. The study follows as closely as possible the DTI methodology which was used to identify such concentrations of economic activity with particular attention directed towards the thirteen clusters identified by the DTI. The paper concludes with some remarks of the general problem of identifying the existence of industrial clusters.