Programació del seminari: Any 2019
- Dimarts, 10 de desembre de 2019. Hora: 12:00
Jaume Barceló i Lídia Montero, Department d'Estadística i Investigació Operativa, UPC
- Divendres, 27 de setembre de 2019. Hora: 12:00
Massimiliano Giacalone, Department of Economics and Statistics, University of Naples
Lp–norm estimators for linear regression models
- Dilluns, 17 de juny de 2019. Hora: 12:00
Melina Castro, Universidad de Costa Rica
- Dimarts, 18 de juny de 2019. Hora: 12:00
Aleix Ruiz de Villa, Barcelona
Introducció a la inferència causal
- Divendres, 15 de març de 2019. Hora: 12:30
David Moriña Soler, Barcelona Graduate School of Mathematics (BGSMath) – Departament de Matemàtiques, Universitat Autònoma de Barcelona (UAB)
Intervention analysis for low count time series with applications in public health
- Divendres, 22 de febrer de 2019. Hora: 12:00
C.F. Jeff Wu, Georgia Institute of Technology, Atlanta, Georgia, USA.
Quality improvement: from autos and chips to nano and bio
- Dimarts, 22 de gener de 2019. Hora: 12:30
Sharon-Lise Normand, Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA.
Statistical Approaches to Health Care Quality Assessments
El uso de datos TIC en la modelización y calibración de redes urbanas. Aplicación de datos de seguimiento GPS y datos de telefonía móvil al modelo metropolitano de Barcelona
CONVIDATS: Lídia Montero i Jaume Barceló
IDIOMA: Castellà
LLOC: Edifici C5, Aula C5016, Campus Nord, UPC
DATA: Dimarts, 10 de desembre de 2019. Hora: 12:00
RESUM:
Actualmente hay diversas compañías que proporcionan mediciones de sistema de posicionamiento global (GPS) procesadas o sin procesar generadas por flotas de vehículos comerciales, aplicaciones de Internet o compañías de automóviles. Otras empresas proporcionan matrices de movilidad origen-destino obtenidas a partir del procesado de datos de telefonía móvil. Las fuentes de datos procedentes de las nuevas tecnologías están entrando en el sector del transporte con fuerza y todavía es objeto de estudio los pros y contras que aportan. El objetivo del seminario es profundizar nuestra comprensión de la aplicabilidad de los datos TIC (tecnologías de la información y la comunicación) en el modelado del transporte. Desafortunadamente, los datos reales a menudo contienen ruido, incertidumbre, errores, redundancias o incluso información irrelevante. Se obtendrían modelos erróneos cuando se construyan sobre datos incorrectos o incompletos. Es por eso que el preprocesamiento es uno de los pasos más críticos en el análisis de datos. Sin embargo, el preprocesamiento no se ha sistematizado adecuadamente, por lo que, a título de ejemplo, se ilustran los pasos necesarios para el preprocesamiento de datos de rastreo GPS y se presenta una propuesta de sistematización. Además, el nivel de agregación está en la ubicación del punto de referencia para posiciones GPS de baja latencia a lo largo de la trayectoria del viaje. Se aborda la fiabilidad del tiempo de viaje en las rutas OD, y cómo en las redes urbanas densas se presentan múltiples posibilidades para la elección de ruta, y los datos de las nuevas tecnologías ofrecen una oportunidad para comprender el comportamiento de elección de ruta.
SOBRE ELS AUTORS:
Jaume Barceló. De 1986 a 2014 catedràtic del Departament d’Estadística i Investigació Operativa de la UPC i des de 2014 Professor Emèrit. La seva recerca ha estat orientada a l’estudi dels models de transport i les tècniques d’optimització i simulació pel seu tractament. Actualment, en col·laboració amb l’empresa alemanya PTV Group, treballa sobre l’explotació de les dades TIC per al seu ús als models de transport .
Lídia Montero. Professora del Departament d’Estadística i Investigació Operativa de la UPC. Doctora en Informàtica per la UPC. Experiència professional en Modelització de Transport i Anàlisi de la Demanda de Transport. Recerca en filtratge de Kalman per l’estimació de matrius origen-destinació dinàmiques i Anàlisi de Dades TIC per aplicacions de transport.
Lp–norm estimators for linear regression models
CONVIDAT: Massimiliano Giacalone
IDIOMA: Anglès
LLOC: Edifici C5, Aula C5016, Campus Nord, UPC
DATA: Divendres, 27 de setembre de 2019. Hora: 12:00
RESUM: Introduction to Exponential Power Distribution (E.P.D.) and Lp–norm estimators. Comparison of the Lp–norm estimators with the Least Squares method (L2) by simulation studies, introducing a linear regression model in the case of multicollinearity. A comparative analysis applying three estimation methods to evaluate the empirical distribution of the regression parameters is considered. A Generalized Error Distribution Copula-based method for portfolios risk assessment. Comparison of E.P.D with Gaussian distribution to financial markets and examples. Final test to all seminar topics.
SOBRE L'AUTOR: Current position: Researcher (SECS-S/01, Statistics) at University of Naples “Federico II” Department of Economics and Statistics – Monte S. Angelo Complex, Via Cinthia, 80126 Naples. Graduate in “Statistics and Economics Sciences”, magna cum laude (Faculty of Economics – University of Palermo) he received his PhD in “Computational Statistics and Applications” from the University of Naples “Federico II”, Department of Mathematics and Statistics (supervisor Prof. N. C. Lauro). During the PhD, he was "visiting scholar" of Prof. A.H. Money in Henley Management College. His research area encompasses the following subjects: Norm-p linear and nonlinear regression - Multidimensional Data Analysis - Big Data - Permutation Tests – Control Charts and Data Quality Control – Applications of Statistics in Economy and in Justice. Local component of the research groups funded by the University of Naples “Federico II” and co-financed by the relevant Ministry, he attended many Statistical Conferences organized by various national and international institutions, presenting numerous communications and papers. Membership of “Società Italiana di Statistica”, “International Association for Statistical Computing”, “International Biometric Society” and “Applied Statistical Association”, he gained over the years considerable teaching experience as Adjunct Professor of “Statistics”, “Economics Statistics”, “Quality Control” , “Probability”, “Statistical Inference”, “Medical Statistics”, at various Italian Universities (Bologna, Naples “Federico II”, Palermo, Catanzaro “Magna Graecia”, Cosenza ”University of Calabria”, Messina, Catania). He is author of about ninety published works concerning Methodological, Social and Economic Statistics.
Integrating Environmental Health characteristics for Health Techonology Assesstment. A systemic approach. The case of the ecosystem in Drake bay, Puntarenas, Costa Rica.
CONVIDAT: Milena Castro
IDIOMA: Castellà
LLOC: Edifici B6, Sala d'Actes Manuel Martí Recober, Campus Nord, UPC
DATA: Dilluns, 17 de Juny de 2019. Hora: 12:00
RESUM: Policy making for environmental health implies consideration of a variety of indicators proposed by theWorld Health Organization and started by Centro de Investigación sobre el Síndrome del Aceite Tóxico y Enfermedades Raras (CISATER). Different observational perspectives can be identified with air, radiation, water, soil, residuals, sanitation, noise, traffic accidents, food safety, infraestructure, ocupational conditions, chemical emergencies and polluted areas. These spatial characteristics can be contrasted with longitudinal community observations of the socio-economic dynamics. However, challenges arise when data available is heterogeneous as comes from a collection of sources of information. A complex model can be defined when integrating more than one conceptual dimension. Dimensions can be specified according to observational techniques: survey data for a socioeconomical and epidemiological characterization of the population, environmental data based on analytic screening of water sources, and clinical epidemiology data can be obtained, in order to elaborate a systemic approach using Markov model simulation. The response of the model is related to the quality of the environment, to identify community development strategies according to its potential and needs satisfaction, like food safety. Model specification allows evaluation of technologies to be implemented at a populational level. Bio-Sand filters were designed and an experimental observation was undertaken with a family in Drake. A decrease in Escherichia coli was observed, but termotolerant coliforms had an increase, after comparing before and after bio-sand filtered water samples from Drake’s main basins. Evidencing a health policy for Drake’s ecosystem implies overtaking microbiological assessments. How an aqueduct should be developed for a population living around areas under forestal and water conservation? Nowadays, this is a relevant research question for Drake’s ecosystem, where biodiversity and water resources represent an important component of its turism based economy.
SOBRE L'AUTOR: : Milena Castro actualment treballa a la Universidad de Costa Rica. És doctora en Bioestadística per la universitat de Leicester, United Kingdom, i la seva formació també abraça l'estadística, l'epidimiologia clínica i la filosofia. També ha participat en diversos projectes tant d'investigació com d'acció social.
Introducció a la inferència causal
CONVIDAT: Aleix Ruiz de Villa
IDIOMA: Català
LLOC: Seminari EIO, ETSEIB (Edifici Eng. Industrial), Planta 6, Campus Sud, Universitat Politècnica de Catalunya, Avda. Diagonal, 647, 08028, Barcelona
DATA: Dimarts, 18 de Juny de 2019. Hora: 12:00
RESUM: Una de les parts més importants en l'anàlisi de dades és estimar quin efecte han tingut certes decisions. La manera més eficient és dur a terme experiments (Randomized Controlled Trials). Tot i així, aquests poden esdevenir molt costosos, no ètics o inviables. A més a més, moltes vegades ja tenim dades i en voldríem treure algun profit. Malauradament, l'anàlisi directa de dades no experimentals pot portar, inclús en casos molt senzills, a conclusions errònies o culs de sac. Un exemple destacat n'és la paradoxa de Simpson. Alguns d'aquests problemes es poden adreçar si s'inclouen elements de causalitat en l'anàlisi. La causalitat ha sigut subjecte d'estudi pels filòsofs des de fa segles. Als anys 80 es va començar a formalitzar des del punt de vista estadístic. Actualment la modelització de la causalitat té tres fonts científiques diferents: computacionals, biomèdiques i economètriques. En aquesta xerrada veurem alguns exemples on modelitzant la causalitat s'obtenen conclusions força diferents a quan no es fa servir. També veurem quan és necessària, quins riscos té i quin tipus de llenguatge i eines en fa ús.
SOBRE L'AUTOR: Aleix Ruiz de Villa és doctor en matemàtiques per la UAB, ex director de data science de LaVanguardia.com, SCRM (responsables de l'app de mòbil en Lidl) i Onna. Fundador del Barcelona Data Science and Machine Learning Meetup (2014) i cofundador del grup d'usuaris de R de Barcelona (2011-2017).
Intervention analysis for low count time series with applications in public health
CONVIDAT: David Moriña Soler
IDIOMA: Català
LLOC: Seminari EIO, ETSEIB (Edifici Eng. Industrial), Planta 6, Campus Sud, Universitat Politècnica de Catalunya, Avda. Diagonal, 647, 08028, Barcelona
DATA: Divendres, 15 de Març de 2019. Hora: 12:30
RESUM: It is common in many fields to be interested in the evaluation of the impact of an intervention over a particular phenomenon. In the context of classical time series analysis a possible choice might be intervention analysis, but there is no analogous methodology developed for low count time series. In this talk, we will introduce a modified INAR model that allows to quantify the effect of an intervention and is also capable of taking into account possible trends or seasonal behaviour. Several examples of application in different real and simulated contexts will also be discussed.
SOBRE L'AUTOR: David Moriña holds a PhD in Mathematics obtained at Autònoma University in 2013. His area of interest is focused in mathematical modelling applied to health sciences, especially in the handling and analysis of longitudinal data, specifically time series data. He has broad experience in the design, development and analysis of clinical trials and epidemiological studies-working in several research centres, including the Technological Center in Nutrition and Health (CTNS), the Centre for Research in Environmental Epidemiology (CREAL) and the Catalan Institute of Oncology, developing new models for cancer research. He joined BGSMath - UAB in 2018, working on the development of new mathematical and statistical models with applications to cancer epidemiology. For more information on his research, see https://bgsmath.cat/people/?person=david-morina-soler
Quality improvement: from autos and chips to nano and bio
CONVIDAT: C.F. Jeff Wu
IDIOMA: Anglès
LLOC: Seminari EIO, ETSEIB (Edifici Eng. Industrial), Planta 6, Campus Sud, Universitat Politècnica de Catalunya, Avda. Diagonal, 647, 08028, Barcelona
DATA: Divendres, 22 de febrer de 2019. Hora: 12:00
RESUM: Quality improvement (QI) has a glorious history, starting from Shewhart’s path-breaking work on statistical process control to Deming’s high-impact work on quality management. Statistical concepts and tools played a key role in such work. As the applications became more sophisticated, elaborate statistical methods were required to tackle the problems. In the last three decades, QI has seen more use of experimental design and analysis, particularly the methodology of robust parameter design (RPD). I will first review some major ideas in RPD, focusing on its engineering origin and statistical methodology. I will then discuss more recent work that expands the original approach, including the use of feedback control and operating window. To have an effective solution, the subject matter knowledge often needs to be incorporated. Techniques for fusing data with knowledge will be presented. For advanced manufacturing and high-tech applications, there are new challenges and possible paradigm shift posed by three features: large varieties, small volume and high added value. I will speculate on some new directions and technical development. Throughout the talk, the ideas will be illustrated with real examples, ranging from the traditional (autos and chips) to the modern (nano and bio).
SOBRE L'AUTOR: C.F. Jeff Wu is Professor and Coca Cola Chair in Engineering Statistics at the School of Industrial and Systems Engineering, Georgia Institute of Technology. He was the first academic statistician elected to the National Academy of Engineering (2004); also a Member (Academician) of Academia Sinica (2000). A Fellow of American Society for Quality, Institute of Mathematical Statistics, of INFORMS, and American Statistical Association. He received the COPSS (Committee of Presidents of Statistical Societies) Presidents’ Award in 1987, the COPSS Fisher Lecture Award in 2011, the Deming Lecture Award in 2012, the inaugural Akaike Memorial Lecture Award in 2016, the George Box Medal from Enbis in 2017, and numerous other awards and honors. He has published more than 175 research articles and supervised 48 Ph.D.'s. He has published two books "Experiments: Planning, Analysis, and Parameter Design Optimization" (with Hamada) and “A Modern Theory of Factorial Designs” (with Mukerjee).
Statistical Approaches to Health Care Quality Assessments
CONVIDAT: Sharon-Lise Normand
IDIOMA: Anglès
LLOC: Edifici B6, Sala d'Actes Manuel Martí, Campus Nord, UPC (veure mapa)
DATA: Dimarts, 22 de Gener de 2019. Hora: 12:30
RESUM: Health plan, hospital, and physician quality assessments are ubiquitous in the U.S. Hospital assessments in particular are used for licensure, maintenance, and some assessments are the basis for modification of hospital payments. For instance, in 2017 the U.S. federal government withheld $528 million from 2597 hospitals as part of the Hospital Readmissions Reduction program described in the Affordable Care Act. This talk will describe the key statistical challenges in determining whether a hospital has higher than "expected outcome" including defining the "treatment", determining the counterfactual outcome, characterizing the role of unmeasured confounders, and accounting for data sparsity and uncertainty.
SOBRE L'AUTOR: Sharon-Lise Normand, Ph.D., is S. James Adelstein Professor of Health Care Policy (Biostatistics) in the Department of Health Care Policy at Harvard Medical School and Professor in the Department of Biostatistics at the Harvard TH Chan School of Public Health. Dr. Normand’s research focuses on the development of statistical methods for health services and outcomes research, primarily using Bayesian approaches, including the evaluation of medical devices in randomized and non-randomized settings for pre- and post-market assessments, causal inference, provider profiling, evidence synthesis, item response theory, and latent variables analyses. Her application areas include cardiovascular disease, severe mental illness, medical device safety and effectiveness, and medical technology diffusion. She earned her Ph.D. in Biostatistics from the University of Toronto, holds a Master of Science as well as a Bachelor of Science degrees in Statistics from the University of Western Ontario, and completed a post-doctoral fellowship in Health Care Policy at Harvard Medical School.
Comparteix: