以下はCFE-CMStatistics 2025 で面白かった/見たかった公演備忘録です
CFE-CMStatistics 2025 講演備忘録
1. 時系列モデル・要因分析 (Factor & VAR Models)
発表テーマ（Dynamic factor stochastic volatility in mean model）とも関連が深い, 動的要因モデルやVARモデルの理論・応用に関するセッション群です. 高次元データに対する新しいアプローチが多く見られました.
- Factor inference under common components in volatility
- Identification and estimation of dynamic factor models
- Bayesian dynamic factor models for high-dimensional matrix-valued time series
- High-dimensional dynamic factor models: A selective survey
- Flexible priors and restrictions for structural vector autoregressions
- Forecasting with time-varying order-invariant structural vector autoregressions
- Time-varying global VARs with application to interconnectedness, structural analysis, and nowcasting
- Generalized impulse responses, multi-horizon projections, and causal mediation analysis in macroeconomics and finance
- Dimension reduction in VAR models via informative lag selection
2. 計算効率と推定アルゴリズム (Computation & Algorithms)
大規模データや複雑なモデルの推定を可能にする, MCMCやガウス過程の新しい計算手法に注目しました.
- A modified algorithm for MCMC in large dimensional models
- Normalizing flows for posterior estimation under intractable likelihoods, with applications in astrostatistics
- Stochastic gradient MCMC for massive geostatistical data
- Vecchia-inducing-points full-scale approximations for Gaussian processes
- Laplace approximations for Gaussian process and mixed effects quantile regression
- Fast generalized spatial multilevel blockNNGP modelling
3. モデル選択とロバスト性 (Selection & Robustness)
ベイズ推定における変数選択, 外れ値への対応, モデルの平均化など, モデルの構造を決定したり, データのノイズに対する頑健性を確保したりする手法です.
- Bayesian outlier detection for matrix-variate models
- Bayesian group variable selection via penalized credible region
- Bayesian model averaging in causal instrumental variable models
- Choice of number of factors and clusters in Bayesian clustering factor models
- Bayesian nonparametric models with BART components
4. 応用と特殊なデータへの対応 (Applications & Specialized Data)
計数データ, 構成データ, 極値など, 特殊な性質を持つデータや, 特定の応用分野に焦点を当てたモデルです.
- Combining GARCH-MIDAS forecasts of US state-level volatility: The role of local and global EPU indices
- Mixture of state space models for compositional data with an application to urban mobility analysis
- Higher-order integer autoregression for count time series
- Dynamic matrix factor model for counts data
- Co-extremal shocks and VAR analysis
- A zero-inflated Poisson latent position cluster model
- Forecast reconciliation and multivariate GARCH
- A regularized regression approach to global minimum variance allocation
- Unified Bayesian nonparametric framework for ordinal, survival, and density regression using complementary log-log link
- IRF and nowcasting with functional approaches and mixed-frequency data
- Distributional outcome regression via quantile functions
]]>