基于商拓扑结构的序列构成和预测

Constituting and Forecasting of Series Based on Quotient Topology

对多变量时间序列进行分析有利于更好地了解各时间序列的特性。根据相关性的时间序列在商空间模型中,可依据信息相关性,该文综合利用多个相关序列提供的信息对其中一个序列进行了预测,通过商空间理论的分解和合成法减小信息不完备产生的影响,从而获得更多准确信息和规则。

Researching these time-series which are interdependent as integration, namely multi-variable time-series analysis, the properties of these time-series can be realized better. For the forecasting question of time-series, in this paper, the relative time-series are dealt synthetically. In quotient space model, based on relativity of information, in order to reduce the impact of the incomplete information, the decomposition and synthesis method of quotient space theory is introduced to obtain more accurate information and rules. The result of experiments shows that the method is effective. Therefore, the solution to the problems is undoubtedly a reasonable approach.