基于聚类分析与时序算法的玉米测产模型

Corn Output Measuring Model Based on Cluster Analysis and Time Series Algorithm

以每公顷玉米产量数据序列为分析处理对象,对玉米产量的时间序列进行了预测分析。结合线性回归分析技术和聚类分析理论,采用分层聚类算法与AR时序算法相融合的方法,探明影响玉米产量的主要因素,确定分层聚类方案,建立一个基于聚类分析的玉米产量AR时序模型,并对2005—2009年的玉米产量进行了预测。聚类分析后模型预测结果的误差值大多数<5%。

Taking corn output data of per hectare series as analysis and processing object,the study on the predictive parsing of corn output time series was carried on.Combing linear regression analysis technique and cluster analysis theory and using the methods of hierarchical clustering algorithms and timing algorithms AR fusion,major factors affecting corn output were ascertained,hierarchical clustering scheme was determined,a corn output AR time series model based on cluster analysis was established,and the 2005 to 2009 corn output was predicted.the error of most model prediction results after clustering analysis is less than 5%.