基于大数据挖掘的橡胶园耕地地力评价  

Evaluation of Land Productivity in Rubber Plantations Based on Data Mining

作　　者：彭金莲[1] 唐朝胜[1] 蒋菊生[2] 赵春梅[3] PENG Jinlian;TANG Zhaosheng;JIANG Jusheng;ZHAO Chunmei(School of Computer Science&Cyberspace Security,Hainan University,Haikou,Hainan 570228,China;Hainan State Farms Academy of Sciences,Haikou,Hainan 570206,China;Rubber Research Institute,CATAS,Danzhou,Hainan 571737,China)

机构地区：[1]海南大学计算机与网络空间安全学院,海口570228 [2]海南省农垦科学院,海口570206 [3]中国热带农业科学院橡胶研究所,海口571101

出　　处：《热带生物学报》2019年第4期380-386,共7页Journal of Tropical Biology

基　　金：海南省自然科学基金项目(20156246)

摘　　要：为了客观评价海南垦区胶园耕地的地力,笔者使用K-Means,Two-Step,Kohonen和TwoStep-AS等4种聚类算法,对海南农垦胶园测土调查数据进行大数据分析,根据轮廓系数(Silhouette Coefficient)的大小,确定TwoStep-AS聚类模型对海南农垦橡胶园耕地地力指标进行聚类分析,经过反复迭代,最终将海南农垦橡胶园耕地划分为3种类型,并筛选出5个对海南垦区胶园耕地类型有影响的地力指标,即年平均降雨量、有效土层厚度、全氮、北纬和成土母质。The data of soil survey in rubber plantations in Hainan State Farms was analyzed to evaluate the land productivity in the rubber plantations by using four clustering algorithms(K-Means, Two-Step, Kohonen and TwoStep-AS). According to the Silhouette coefficients, TwoStep-AS clustering models were established for cluster analysis of the indexes of the land productivity in the Hainan State Farm rubber plantations. After repeated iterations, the land in rubber plantations of Hainan State Farms was divided into three types, and the indexes affecting the land type of the rubber plantations were selected, including annual average rainfall, effective soil thickness, total nitrogen, north latitude, and soil-forming parent material.

关 键 词：橡胶园 大数据挖掘 地力评价 轮廓系数 

分 类 号：S-3[农业科学]