【作者】 张元；

【导师】 韩燮；

【作者基本信息】 华北工学院 ， 计算机应用技术， 2004， 硕士

【摘要】 随着数据库技术的飞速发展，数据库不仅在数量上快速增长，规模也越来越大。在许多现实的数据库中，数据常常与时间有关。按时间顺序取得的一系列观测值被称为时间序列数据。时间序列数据是一种复杂的数据对象，在社会生活中的各个领域广泛存在着大量的时间序列数据有待于进一步的分析和处理。时间序列的数据挖掘对人类社会、科技和经济的发展有重大意义。目前时间序列的数据挖掘是数据挖掘的重要研究热点之一。本论文首先介绍了本课题的背景、研究内容和研究意义，论述了数据挖掘、时序数据挖掘的研究现状，包括它们的研究内容、常用方法以及当前的研究重点和发展方向。其次，研究了常用的时序数据趋势分析模型，并对它们的推理过程和适用性进行了详细的阐述。指出了现有方法的局限性，提出了将智能决策支持系统引入时序数据趋势分析必要性和可行性。第三，分析了基于智能决策支持的时序数据趋势分析系统的设计思想、设计目标和总体设计框架。第四，详细从人机界面、知识库、模型库和自学习机制方面设计了一个基于智能决策支持的时序数据趋势分析系统。最后，通过对一个实际时间序列的分析，给出了实验结果，验证了该系统的有效性和可行性。更多还原

【Abstract】 With the development of database technology, the databases have increased not only in quantities but also in scales. The data always are related to time in many practical databases. Time series data is a series of observational data according to temporal sequence. Time series is very important and complicated data object. There are a lot of time series need to be further analyzed and processed in the all kinds of areas in society. Time series data mining is of great importance to development of society, science and technology and economy to the people. Research on time series data mining is one of important hot spots of data mining.Firstly, after the background, content and intention of this paper are clarified; status in quo of data mining and time series data mining are discussed, including research content, common ways, current research emphases and development trend.Secondly, time series trend analysis models in common use are introduced, whose illative process and applicability are also expatiated. And shortcomings of the existing ways are pointed out; necessity and feasibility that intelligent decision support system is introduced time series trend analysis are put forward.Thirdly, design thinking, design objective and general frame of time series trend analysis system based on intelligent decision support are analyzed.Fourthly, a time series trend analysis system based intelligent decision system is designed in detail from the point of human-computer interface, knowledge base, model base and machine learning.Lastly, validity and feasibility of time series trend analysis system based on intelligent decision support are proved by analyzing an actual time series and the experimental result.更多还原