河北省粮食产量预测影响因素关联挖掘分析

Correlation analysis of influencing factors of grain yield forecast in Hebei province

粮食产量受多种因素影响,并且波动性较大,本研究以河北粮食产量为例,利用网络爬虫技术获取众多农业网站中影响粮食产量的数据,针对数据存在的类型复杂、冗余、缺失问题,对其进行清洗、集成,利用拉格朗日插值法对缺失数据进行补充。在此基础上利用灰色关联和Lasso回归模型完成特征值的筛选和影响因子权重分析,为充分考虑历史信息验证关联分析的准确性,提高模型的自适应性和容错能力,建立了灰色预测与神经网络组合的预测模型,通过Echarts可视化技术进行数据展示。实验结果表明,利用网络爬虫搜集到的农业统计数据通过数据处理和关联挖掘分析,使粮食产量预测的相对误差在0.03%左右,提高了预测精度。

Food production is influenced by many factors and highly volatile.Taking the food production in Hebei as an example,the web crawler technology was used for getting data that affects grain production in various agricultural sites.In view of the data problems of complex types,redundancy,and data missing,the data was cleaned and integrated,and the method of Lagrange interpolation was adopted to supplement the missing data.The eigenvalue filtering and weight analysis of impact factors were done by using the grey correlation and Lasso regression model.To fully consider the accuracy of correlation analysis for historical information,improve the adaptability of the model and fault tolerance,the combination forecast model of gray prediction and neural network was established,and the data was displayed through Echarts visualization technology.The experimental results showed that the relative error of grain yield prediction was about 0.03%through data processing and correlation mining analysis for agricultural statistical data collected by web crawlers,which improved the prediction accuracy.