Multivariate models and inference

经典的多元统计和时间序列统计方法假设观测值的分布呈钟形曲线。多元统计包括分析数据的方法，这些数据具有许多不同的数值测量或变量的形式，适用于许多不同的主题或情况；感兴趣的一个推论是不同变量之间的关联或关系的数量。此外，当变量与金融和保险风险相关时，研究一种称为尾部依赖的依赖形式非常重要。时间序列包括用于分析随时间顺序获得的主要变量和可能同时测量的其他协变量的测量形式的数据的方法；我们感兴趣的一个推论是预测或预测时间序列的未来值。

Classical statistical methods for multivariate statistics and time series have an assumption that the distributions of observations has a bell-shaped curve. Multivariate statistics consists of methods for analyzing data that has the form of many different numerical measurements or variables for many different subjects or cases; one inference of interest is the amount of association or relationship among different variables. Also a form of dependence called tail dependence is very important to study when the variables are related to financial and insurance risks. Time series consists of methods for analyzing data of the form of measurements of a main variable obtained sequentially over time and possibly other co-variables that are measured simultaneously; one inference of interest is to predict or forecast future values of the time series.