quinta-feira, março 19, 2015

XVI Escola de Séries Temporais e Econometria


XVI Escola de Séries Temporais e Econometria


Marcelo Fernandes (FGV-SP)

Dani Gamerman (UFRJ)

S.J. Koopman (VU Amsterdam)

Alexandra Schmidt (UFRJ)

Ruey Tsay (Booth School of Business - Chicago)

Brani Vidakovic (GeorgiaTech)

Mauricio Zevallos (UNICAMP)

Wilfredo Palma (UC Chile)

ST1: High Dimensional Time Series
Chairman: Flávio Ziegelmann
Siem J. Koopman
Marcelo Medeiros
Guilherme Moura

ST2: High Dimension Volatility Models
Chairman: Hedibert Lopes
Ruey Tsay
Hedibert Lopes
André Portela

ST3: Discrete-Valued Time Series
Chairman: Valdério Reisen
Wilfredo Palma
Glaura Franco
Klaus Vasconcelos

ST4: Regularized Regression and Classification
Chairman: Marcelo Fernandes
Brani Vidakovic
Eduardo Mendes
Aluisio Pinheiro


 State Space Models: Theory, Methods and Applications

 David Stoffer (U. Pittsburgh)

The state space model (SSM) or the hidden Markov Model (HMM) is a very general model that subsumes a whole class of special cases of interest in much the same way that regression does.  For example, the linear Gaussian model includes such varied models ARMA models as well as smoothing splines.  Nonlinear state space models are used in finance, in particular, stochastic volatility as well as computational ecology to study population dynamics and to track animals.  While inference for linear state space models is fairly simple using numerical optimization based on derivatives, inference in the nonlinear case is difficult and relies on derivative free numerical optimization. I will introduce the basic model along with some theory and applications.  I will then introduce a variety of nonlinear models, and discuss applications and inference for these models based on Monte Carlo methods including MCEM, Metropolis-Hastings, and particle methods.


T1 - Time-varying Copulas - Flávio Ziegelmann

T2 - An Introduction to Singular Spectrum Analysis - Paulo Canas Rodrigues

T3 - Bayesian Regularization - Hedibert Lopes

T4 - Resampling Techniques for Nonstationary Time Series - Jacek Leskow