[ABE-L] Bayesian Statistical Learning in Econometrics
Hedibert Lopes
hedibert em gmail.com
Ter Abr 10 17:09:53 -03 2018
Boa tarde povo.
Gostaria de divulgar um curso em nivel de doutorado que lecionarei nas
proximas varias semanas aqui no Insper. Detalhes abaixo. Interessados
entrar em contato comigo para verificar viabilidade.
http://hedibert.org/current-teaching/#tab-BayesianEconometrics-2018
Obrigado e abracos,
Hedibert
Objective
The end of the course goal is to allow the student to critically decide
between a Bayesian, a frequentist or Bayesian-frequentist compromise when
facing real world problems in the fields of micro- and macro-econometrics
and finance, as well as in quantitative marketing, strategy and business
administration. With this end in mind, we will visit well known Bayesian
issues, such as prior specification and model comparison and model
averaging, but also study regularization via Bayesian LASSO, Spike-and-Slab
and related schemes, “small n, large p” issues, Bayesian statistical
learning via additive regression trees, random forests, large-scale VAR and
(dynamic) factor models.
Course description
Basic ingredients: prior, posterior, and predictive distributions,
sequential Bayes, conjugate analysis, exchangeability, principles of data
reduction and decision theory. Model criticism: Bayes factor, computing
marginal likelihoods, Savage-Dickey ratio, reversible jump MCMC, Bayesian
model averaging and deviance information criterion. Modern computation via
(Markov chain) Monte Carlo methods: Monte Carlo integration,
sampling-importance resampling, Gibbs sampler, Metropolis-Hastings
algorithms. Mixture models, Hierarchical models, Bayesian regularization,
Instrumental variables modeling, Large-scale (sparse) factor modeling,
Bayesian additive regression trees (BART) and related topics, Dynamic
models, Sequential Monte Carlo algorithms, Bayesian methods in
microeconometrics, macroeconometrics, marketing and finance.
Part I Bayesian ingredients
Inference: likelihood, prior, predictive and posterior distributions
Model criticism: Marginal likelihoods, Bayes factor, model averaging and
decision theory
Computation: An introduction (Markov chain and sequencial) Monte Carlo
methods
Part II Multivariate models
Large-scale vector autoregressive models
Factor models and other dimension reduction models
Time-varying high-dimensional covariance models
Part III Modern Bayesian statistical learning
Mixture models and the Dirichlet process: handling non-Gaussian models
Regularization: sparsity via shrinkage and variable selection
Large vector-autoregressive and factor models: combining sparsity and
parsimony
Classification and support vector machines
Regression trees and random forests
Latent Dirichlet allocation: Text as data, text mining
-------------- Próxima Parte ----------
Um anexo em HTML foi limpo...
URL: <https://lists.ime.usp.br/archives/abe/attachments/20180410/5abe1b56/attachment.html>
Mais detalhes sobre a lista de discussão abe