大阪大学　数理・データ科学教育研究センター
Center for Mathematical Modeling and Data Science,Osaka University

Overview of copulas and Bayesian ridge estimator

江村 剛志　教授（統計数理研究所）

1. ホーム
2. セミナー・集中講義
3. セミナー
4. Overview of copulas and Bayesian ridge estimator

大阪大学 データ科学セミナーシリーズ 第59回

Overview of copulas and Bayesian ridge estimator

江村 剛志　教授（統計数理研究所）

Copulas are statistical tools for modeling dependence between variables. I first provide a brief introduction to copulas and vine copulas. Then, I talk about their applications to Bayesian ridge estimators for linear models. In these applications, we propose vine copulas as the prior for regression coefficients. We show that the vine-copula prior gives a more accurate posterior estimator compared to the traditional multivariate normal prior. The proposed model is illustrated by fitting the model (Y=cars' CO2 emissions, X1=engine sizes, X2=car weights, X3=interaction).