Bayesian High-Dimensional Time series models with applications to Macroeconomic and Financial data

这个项目的重点是开发灵活的贝叶斯方法来建模时间序列数据。具体来说，我们致力于从贝叶斯的角度发展推理方法，然后是经济和金融中出现的高维时间序列数据的预测技术。例如，投资组合成分的回报向量，或不同国家的信用违约掉期数据向量。关于方法问题，该项目的重点是开发针对低秩线性模型的贝叶斯方法，如先验规范和有效参数估计。这样的模型预计将非常适合于经济和金融数据，因为这些数据通常具有比实际观察到的低得多的维度结构。贝叶斯方法低秩线性建模将使模型参数的自适应确定是难以估计的经典方法。

This project focuses on the development of flexible Bayesian methods for modelling time series data. Specifically, we work on developing inference methodology from a Bayesian perspective, and then forecasting techniques for high dimensional time series data arising in economics and finance. For example, the vector of returns of portfolio constituents, or a vector of credit default swaps data for different nations.Regarding the methodological issue, the focus of the project is on developing Bayesian methods specific to low-rank linear models, such as prior specification and efficient parameter estimation. Such models are expected to be well-suited to economics and finance data, as such data often possesses structure with dimension much lower than that what is actually observed. A Bayesian approach to low-rank linear modelling will enable adaptive determination of model parameters that are hard to estimate with classical methods.