Seminario de Estadística
2016-10-14
12:00hrs.
12:00hrs.
Carlos Díaz Ávalos. Instituto de Investigaciones en Matemáticas Aplicadas y en Sistemas - Unam, México
Estadística espacial: Contestando preguntas relevantes en ciencias ambientales y de la salud
Sala 2, Facultad de Matemáticas
Abstract:
Estadística espacial: Contestando preguntas relevantes en ciencias ambientales y de la salud
Sala 2, Facultad de Matemáticas
Abstract:
La estadística espacial es una rama que cobró auge a finales de la década de los 80, cuando se desató una procupación mundial por problemas ambientales. Esto dió lugar al desarrollo de métodos aplicables a ciencias ambientales en las que se consideran campos aleatorios de tipo contínuo, discreto y puntual. Actualmente la mayoría de los países enfrentan graves problemas ambientales entre los que destacan la quema de bosques, la contaminación y problemas epidemiológicos.
En esta charla se presenta una visión global de los métodos de la estadística espacial adecuados para el análisis en los tres tipos de soporte y se muestran tres aplicaciones a datos reales, con una breve reseña de tópicos de investigación aún abiertos.
2016-10-07
12:00hrs.
12:00hrs.
Giovanni Motta. Pontificia Universidad Católica de Chile
Local Polynomials for time-varying correlations: adaptivity versus positivity
Sala 2, Facultad de Matemáticas
Abstract:
In this paper we propose a new nonparametric method to estimate the time-varying correlation between two non-stationary time series. Linear smoothers of the cross-products are based on the same bandwidth for both numerator (covariance) and denominator (variances). This approach guarantees two important properties: the estimated correlation is bounded between minus one and one, and the resulting correlation matrix is positive semi-definite. However, the use of one common bandwidth for both numerator and denominator appears to be restrictive, as the covariance and the variances are in general characterized by different degrees of smoothness. On the other hand, a kernel-type estimator based on different smoothing parameters for numerator and denominator has two drawbacks. First, the ratio between time-varying numerators and denominators is not necessarily bounded between minus one and one; as a consequence, the resulting correlation matrix is not necessarily positive semi-definite. Second, the estimated bandwidths that are optimal for estimating the covariance and the variances are not necessarily optimal for estimating the ratio. The estimator we propose in this paper is based on local smoothing of the sign of the cross-products, which does not require distinguishing between numerator and denominator. Our novel method can be used to estimate the time-varying AR coefficients and time-varying spectra of locally stationary time series.
Local Polynomials for time-varying correlations: adaptivity versus positivity
Sala 2, Facultad de Matemáticas
Abstract:
In this paper we propose a new nonparametric method to estimate the time-varying correlation between two non-stationary time series. Linear smoothers of the cross-products are based on the same bandwidth for both numerator (covariance) and denominator (variances). This approach guarantees two important properties: the estimated correlation is bounded between minus one and one, and the resulting correlation matrix is positive semi-definite. However, the use of one common bandwidth for both numerator and denominator appears to be restrictive, as the covariance and the variances are in general characterized by different degrees of smoothness. On the other hand, a kernel-type estimator based on different smoothing parameters for numerator and denominator has two drawbacks. First, the ratio between time-varying numerators and denominators is not necessarily bounded between minus one and one; as a consequence, the resulting correlation matrix is not necessarily positive semi-definite. Second, the estimated bandwidths that are optimal for estimating the covariance and the variances are not necessarily optimal for estimating the ratio. The estimator we propose in this paper is based on local smoothing of the sign of the cross-products, which does not require distinguishing between numerator and denominator. Our novel method can be used to estimate the time-varying AR coefficients and time-varying spectra of locally stationary time series.
2016-07-07
María Dolores Jiménez-Gamero. Universidad de Sevilla
Penalized Estimation of The Finite Population Distribution Function Using Auxiliary Information
Sala 2 (Víctor Ochsenius) Facultad de Matemáticas a las 12:00 Hrs.
Penalized Estimation of The Finite Population Distribution Function Using Auxiliary Information
Sala 2 (Víctor Ochsenius) Facultad de Matemáticas a las 12:00 Hrs.
2016-07-07
Pedro Jodrá. Universidad de Zaragoza
On The Log-Extended Exponential-Geometric Distribution and Applications in Business Research
Sala 2 (Víctor Ochsenius) Facultad de Matemáticas a las 12:00 Hrs.
On The Log-Extended Exponential-Geometric Distribution and Applications in Business Research
Sala 2 (Víctor Ochsenius) Facultad de Matemáticas a las 12:00 Hrs.
2016-03-28
Eduardo Engel. Facultad de Economía de la Universidad de Chile
Sala 2 de la Facultad de Matemáticas a las 12hs.
Sala 2 de la Facultad de Matemáticas a las 12hs.
2016-03-28
Eduardo Engel. Facultad de Economía de la Universidad de Chile.
Missing Aggregate Dynamics: on The Slow Convergence of Lumpy Adjustment Models
Sala 2 - Facultad de Matemáticas PUC
Missing Aggregate Dynamics: on The Slow Convergence of Lumpy Adjustment Models
Sala 2 - Facultad de Matemáticas PUC
2016-03-18
Paula Brito. Universidad de Porto
Taking Variability in Data Into Account: Symbolic Data Analysis The Special Case of Interval Data
Sala 2- Facultad de Matemáticas 12:00 Hrs.
Taking Variability in Data Into Account: Symbolic Data Analysis The Special Case of Interval Data
Sala 2- Facultad de Matemáticas 12:00 Hrs.
2016-01-22
Hamdi Raissi
Statistical Analysis of The Non Constant Covariance Structure of Time Series.
Sala 2 Facultad de Matemáticas a las 12:00 Hrs.
Statistical Analysis of The Non Constant Covariance Structure of Time Series.
Sala 2 Facultad de Matemáticas a las 12:00 Hrs.
2016-01-15
Mattieu Saumard
Two Applications of Functional Data.
Sala 2 Facultad de Matemáticas a las 12:00 Hrs.
Two Applications of Functional Data.
Sala 2 Facultad de Matemáticas a las 12:00 Hrs.
2016-01-14
Giovanni Motta
Tba
Sala 2 Facultad de Matemáticas a las 15:00 Hrs.
Tba
Sala 2 Facultad de Matemáticas a las 15:00 Hrs.
2016-01-12
Luis Gutierrez
A Bayesian Approach for Statistical Shape Analysis.
Sala 2 - Facultad de Matemáticas UC a las 15:00 Hrs.
A Bayesian Approach for Statistical Shape Analysis.
Sala 2 - Facultad de Matemáticas UC a las 15:00 Hrs.
2016-01-07
Isabelle Beaudry
Improved Inferential Procedures in Respondent-Driven Sampling
Sala 2 - Facultad de Matemáticas a las 15:15 Hrs.
Improved Inferential Procedures in Respondent-Driven Sampling
Sala 2 - Facultad de Matemáticas a las 15:15 Hrs.
2016-01-05
Daniel Turek
The Nimble Project and Efficient Markov Chain Monte Carlo Algorithms
Sala 2 - Facultad de Matemáticas a las 16:00 Hrs.
The Nimble Project and Efficient Markov Chain Monte Carlo Algorithms
Sala 2 - Facultad de Matemáticas a las 16:00 Hrs.
2015-12-21
Heejung Shim. Universidad de Prude
Multi-Scale Methods for Analyses of Functional Phenotypes Arising From High-Throughput Sequencing Assays
Sala 5 de la Facultad de Matemáticas a las 15:00 hrs.
Multi-Scale Methods for Analyses of Functional Phenotypes Arising From High-Throughput Sequencing Assays
Sala 5 de la Facultad de Matemáticas a las 15:00 hrs.
2015-12-18
Catalina Vallejos. Universidad de Warwick/ Instituto Europeo de Bioinformática
Basics: Bayesian Analysis of Single-Cell Sequencing Data
Sala 2 de la Facultad de Matemáticas a las 12:00 Hrs.
Basics: Bayesian Analysis of Single-Cell Sequencing Data
Sala 2 de la Facultad de Matemáticas a las 12:00 Hrs.
2015-06-05
Sandor Baran. University of Debrecen, Faculty of Informatics
Aymptotic Properties of Parameter Estimators in Pickard Models
Sala 2 Facultad de Matemáticas UC 12:00 Hrs.
Aymptotic Properties of Parameter Estimators in Pickard Models
Sala 2 Facultad de Matemáticas UC 12:00 Hrs.
2015-05-15
Felipe Osorio. Instituto de Estadística, Pontificia Universidad Católica de Valparaíso
Análisis Multivariado Usando la Distribución T: Una Aplicación Con Datos de Afp Chilenas
Sala 2 Facultad de Matemáticas
Análisis Multivariado Usando la Distribución T: Una Aplicación Con Datos de Afp Chilenas
Sala 2 Facultad de Matemáticas
2015-04-24
Carolina Casas-Cordero. Pontificia Universidad Católica de Chile, Instituto de Sociología
Estimación Para Áreas Pequeñas: Aplicación Para Las Tasas de Pobreza Comunal en Chile
Sala 2 Facultad de Matemáticas - 12:00 Hrs
Estimación Para Áreas Pequeñas: Aplicación Para Las Tasas de Pobreza Comunal en Chile
Sala 2 Facultad de Matemáticas - 12:00 Hrs
2015-04-17
Cristian Meza. Universidad de Valparaíso, Facultad de Igeniería
Segmentation of Multiple Series Using a Lasso Strategy
Sala 2 Facultad de Matemáticas - 12:00 Hrs.
Segmentation of Multiple Series Using a Lasso Strategy
Sala 2 Facultad de Matemáticas - 12:00 Hrs.
2015-03-27
Christian Berg. Universidad de Copenhague
On Moment Problems - Historical Origins, Significance and Recent Developments
Sala 2 Facultad de Matemáticas - 12:00 Hrs.
On Moment Problems - Historical Origins, Significance and Recent Developments
Sala 2 Facultad de Matemáticas - 12:00 Hrs.
Contáctenos
Edificio Rolando Chuaqui, Campus San Joaquín. Avda. Vicuña Mackenna 4860, Macul, Chile.Teléfono (+56) 95504 4511.
Formulario de Contacto