2017 October : Biostatnet

Seminarios Prof. Douglas Wiens
Con motivo de la visita del profesor Douglas Wiens, profesor emérito de la Universidad de Alberta, Canadá, el nodo Castilla La Mancha-OED ha organizado un seminario con distintas conferencias en Toledo el 27 de octubre y otro en Pamplona el 30 de octubre de 2017. OED SEMINAR UNIVERSITY OF CASTILLA-LA MANCHA TOLEDO – Friday, 27th October 2017 Location: Room 31.14 Building: 31 Campus Fábrica de Armas Toledo Schedule: 12:30 h.  Douglas Wiens (University of Alberta, Canadá) Title:  Robustness of Design: A Survey Abstract:  When an experiment is conducted for purposes which include fitting a particular model to the data, then the ‘optimal’ experimental design is highly dependent upon the model assumptions – linearity of the response function, independence and homoscedasticity of the errors, etc. When these assumptions are violated the design can be far from optimal, and so a more robust approach is called for. We should seek a design which behaves reasonably well over a large class of plausible models. I will review the progress which has been made on such problems, in a variety of experimental and modelling scenarios – prediction, extrapolation, discrimination, survey sampling, dose-response, machine learning, etc. 13:30 h.  Sergio Pozuelo-Campos. Title: Optimal Experimental Design for Different Probability Distributions. Abstract:  In Optimal Experimental Design, we usually assume the normal distribution for the responses, hence the errors have a normal distribution. The Central Limit Theorem ensure us that, if the ­sample size is big enough, this assumption is not a problem, but are not always able to do so. Therefore, the behaviour of the optimal experimental design is approached for others common probabiity distributions commonly used in various areas of knowledge. We compute and compare D-optimal designs for different linear and non-linear models assuming different probability distributions. 13:50 h.   Irene García-Camacha. Title: Efficient algorithms for constructing D-optimal designs for linear and non-linear models in mixture experiments. Abstract: Mixture experiments analyse systems defined over a simplex-shaped experimental region. Mixture design background has mainly been based on a classical design approach. There is not much literature for mixtures in the context of optimal experimental design, specifically for non-linear models. Analytical solutions can only be found on examples where strict assumptions have to be included. They are far from realistic scenarios. In spite of this simplification, numerical methods are needed to construct optimal designs. Under this framework, it is necessary to develop general optimization techniques in order to find optimal solutions for these problems. Two efficient algorithms are proposed in this paper for computing exact D-optimal designs in order to deal with the special nature of mixture experiments. They are based on a multiplicative algorithm and a genetic algorithm. Several examples illustrate the enhanced results achieved by these new methods.  
Seminario “Data Science: de los datos a la toma de decisiones eficiente”
Seminarios organizados por el grupo de Estadística Aplicada del Basque Center for Applied Mathematics (BCAM) y los departamentos de Economía Aplicada III (Econometría y Estadística) y Matemática Aplicada, Estadística e Investigación Operativa de la UPV/EHU. Título: Data Science: De los datos a la toma de decisiones eficiente Jose Miguel Carot Sierra (Universidad Politécnica de Valencia) 27 de Octubre 13:00 BCAM Basque Center for Applied Mathematics Mazarredo 14, 48009 Bilbao
Día de la Estadística de Catalunya 2017
Día de la Estadística a Catalunya 2017 Convocado por la Societat Catalana d'Estadística y organizado por el Servei d'Estadística Aplicada, UAB. Viernes, 6 de octubre de 2017 Más información: enlace
Seminario “Extensiones metodológicas en el ámbito de la Regresión. Aplicaciones en biomedicina”
Seminario “Extensiones metodológicas en el ámbito de la Regresión. Aplicaciones en biomedicina” Celebrado en la facultad de Medicina el día 1 de octubre de 2017. Participaron como docentes miembros del GRIDECMB y del departamento de Estadística, Análisis Matemático y Optimización de la USC. Las ponencias impartidas fueron las siguientes: “Joint Modelling of non-linear multivariate longitudinal and survival data using penalized spline smoothing. A two-stage based approach” “Flexible Bayesian additive joint models with an application to type 1 diabetes research” “Testear modelos de regresión cuantil con covariables en alta dimensión usando proyecciones” “Modelos aditivos generalizados de localización, escala y forma para respuestas bivariadas. Aplicaciones en diabetes” “Flexible Joint Modelling including Functional data. Application in diabetes research”.