基于C4.5决策树算法的农安县农田管理区划分

C4.5 Decision Tree Algorithm Based Division of Farmland Management Zone in Nong'an County

针对吉林省长春市农安县2012年耕地地力调查数据,使用数据挖掘中的C4.5决策树算法(以下简称C4.5算法)对土壤中速效氮、速效磷和速效钾3种主要养分含量进行分析,并根据各养分间差异划分管理区。结果表明,使用C4.5算法对农安县924、1 840、2 756组土壤速效养分数据进行分析得到的模型精度分别为95.81%、95.59%、95.13%,对测试样本的分类精度分别为88.25%、90.50%、91.57%。使用C4.5算法对农安县进行管理区划分准确度高、效率高、易于推广,是一种指导农安县精细农业实施的有效管理区划分方法。

Based on the farmland survey data of Nong＆#39;an county in 2012, C4.5 decision tree algorithm was used to analyze three soil nutrients （nitrogen, phosphorus and potassium﹚, and divide management zone according to the difference between the nutrients. Results showed that the accuracy of the model acquired by C4.5 decision tree algorithm was 95.81%, 95.59%and 95.13% for 924, 1 840, 2 756 soil nutrient data of Nong’an county. The classification accuracy of test data was 88.25%, 90.50% and 91.57%, respectively. The management partition based on C4.5 algorithm was efficient, accurate and easy to popularized, which was an effective management partition method to guide variable fertilization in precision agriculture of Nonga’n county.