引用本文:
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 159次   下载 108 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于随机森林算法的耕地面积预测及影响因素重要性分析——以甘肃省庆阳市为例
王全喜1, 孙鹏举2,3, 刘学录2, 李尚泽2, 高建存4
1.甘肃农业大学 管理学院, 甘肃 兰州 730070;2.甘肃农业大学 资源与环境学院, 甘肃 兰州 730070;3.甘肃省国土资源规划研究院, 甘肃 兰州 730000;4.中国地质大学(武汉) 公共管理学院, 湖北 武汉 430074
摘要:
[目的]分析耕地面积变化影响因素的重要性,以便科学预测耕地资源数量,为保护耕地资源服务。[方法]以属于黄土高原地区的甘肃省庆阳市为例,尝试采用随机森林算法构建耕地面积预测模型,与BP神经网络模型的预测结果进行对比,并对耕地面积变化影响因素重要性进行排序。[结果]随机森林算法预测结果的相对误差和均方根误差均小于BP神经网络的,预测精度高,结果稳定。它预测出2020,2025,2030年的耕地面积分别为4.515×105,4.513×105,4.512×105 hm2,呈现减少的趋势;主要影响因素重要程度排序为:农业机械总动力 > 农业人口 > 地区生产总值 > 固定资产投资额。[结论]随机森林算法适合于耕地面积预测,且能够测度耕地面积变化影响因素的重要程度。
关键词:  耕地面积  随机森林算法  预测  庆阳市
DOI:10.13961/j.cnki.stbctb.2018.05.054
分类号:F301.24
基金项目:甘肃省自然基金项目“生态脆弱区的土地利用与生态安全研究”(GSAN-ZL-2015-045)
Prediction of Cultivated Land Area and Importance of Influencing Factors Based on Random Forest Algorithm—A Case Study of Qingyang City, Gansu Province
WANG Quanxi1, SUN Pengju2,3, LIU Xuelu2, LI Shangze2, GAO Jiancun4
1.College of Management, Gansu Agricultural University, Lanzhou, Gansu 730070, China;2.College of Resources and Environmental Sciences, Gansu Agricultural University, Lanzhou, Gansu 730070, China;3.Land Resources Planning and Research Institute of Gansu Province, Lanzhou, Gansu 730000, China;4.College of Public Management, China University of Geoscience, Wuhan, Hubei 430074, China
Abstract:
[Objective] To analyze the importance of the factors that influence the change of cultivated land area in order to predict the amount of cultivated land area resources, and to service the protection of cultivated land.[Methods] Taking Qingyang City of Gansu Province as a case study, the random forest algorithm was used to construct the prediction model of cultivated land area. The results were compared with those of BP neural network model, and the importance of the factors that influencing cultivated land area change was sorted.[Results] The relative error and root mean square error of the prediction results of the random forest algorithm were smaller than that of BP neural network, and the prediction accuracy was high and the results were stable. The cultivated land area in 2020, 2025 and 2030 was predicted to be 4.515×105, 4.513×105 and 4.512×105 hm2, respectively, showing a decreasing trend. The importance of the main influencing factors was ranked as:agricultural machinery general dynamics > agricultural population > GDP > fixed asset investment.[Conclusion] The random forest algorithm is suitable for the prediction of cultivated land area and can measure the importance of factors that influence the change of cultivated land area.
Key words:  cultivated land area  random forest algorithm  prediction  Qingyang City