ASYMPTOTIC THEORY FOR TIME SERIES ANALYSIS

摘要

An observed sequence which is supposed to vary randomly with time, interacting with (depending on) the past, present and future states, is called a time series. Mathematically we consider this sequence as a realization of a stochastic process (a family of random variables). The statistics for a stochastic process is called time series analysis. Therefore, if we say that the usual statistics is the one for independent samples, we may say that time series analysis is the statistics for dependent samples, and is regarded as a more generalized statistics.