基于大数据的小麦蚜虫发生程度决策树预测分类模型

Decision tree predictive classification model on the occurrence degree of wheat aphids based on big data

小麦蚜虫是危害小麦的主要害虫。其发生程度预测特别是短期预测一直是植物保护领域难以解决的科学问题。传统预测方法通常仅采用温湿度,预测结果与实际发生匹配度不高。基于大数据的理念和数据挖掘技术,通过对2003-2013年小麦蚜虫发生程度与瓢虫、寄生蜂、日最高气压、日照时数等18种变量关系的决策树分析,构建了分类模型。经分析发现,日照时数与小麦蚜虫的发生程度关联度最高,其次是天敌瓢虫。该模型置信度为91.49%,且运行稳健。

Wheat aphids are the main pests of wheat crops. The monitoring and forecasting of their occurrence degree, especially the short-term occurrence degree, is much difficult. Many traditional predictions rely only on temperature and humidity, so the match degree to the actual occurrence value is low. Based on the concept of big data and data mining programs, the predictive classification model was established by means of the decision tree analysis of the relationship between the occurrence degree of aphids and up to 18 variables. It was found out that the duration of sunshine has the highest degree of relevance to the forecasting level of aphids, followed by ladybird. The confidence coefficient of the model that runs steadily in the experiment is 91.49%.