加权零阶局域法在网络舆论趋势预测中的应用

Application of adding-weight zero-rank local-region method on prediction of public opinion on the Internet

采集了某一时段国内某著名网站上针对某一指定话题的帖子,按时间周期构造了一个关于新增帖子数量的时间序列,利用混沌时间序列预测方法中的加权零阶局域预测法预测了该序列的变化趋势.仿真结果表明该预测趋势与实际趋势在短期内是基本一致的;短期预测误差明显小于长期预测误差.

To forecast public opinion trend on the Internet, many notes were collected from a famous website＇s BBS, which came forth in a period of time. A time series of the number of new notes was built according to time cycle. Then the series change and trend were predicted by the adding-weight zerorank local-region method in chaotic time series prediction method. the prediction number was consistent with the fact number in short error was obviously smaller than the long-term error. The simulation result indicated that term, and the short-term prediction