基于滑动窗口的时间序列数据流分析与预测技术研究

数据流管理与分析的研究是目前国际数据库领域的一个研究热点,数据流上回归分析是一项非常必要而有意义的工作.结合数据流的特征,研究了时间序列数据流的回归分析与预测技术,以及一元线性回归方程的聚集技术,给出了时间序列数据流上的一元线性回归分析模型.在此基础上,提出了一种数据流上的预测模型.最后,试验分析展示了研究结果能够有效地产生时间序列数据流的回归模型和实现数据流未来数据的预测.

Study on regression analysis and prediction of time-series data streams using sliding windows

The study on data streams is one of the hot topics among the database circle all over the world recently. The regression analysis of data streams is a necessary and significant task. The authors investigate methods for regression analysis and prediction of time-series data streams. Firstly, a technology of building linear regression model of time-series data streams and the predicting applications of the linear regression model are investigated, Then a technology for aggregating the multiple linear regression equation of time-series data streams. Finally, the results show the methods can effectively come into being regression analysis model of time-series data streams, and fulfill the prediction of future data streams.