基于Kohonen网络和DEA交叉评价耕地利用效率聚类分析  

Cluster analysis of cultivated land use efficiency based on Kohonen network and DEA cross evaluation

作　　者：刘轩[1] 牛海鹏[1] 董国权[1] 刘文锴[2] 

机构地区：[1]河南理工大学测绘学院,河南焦作454000 [2]河南工程学院土木工程学院,河南郑州451191

出　　处：《浙江农业学报》2015年第9期1652-1658,共7页Acta Agriculturae Zhejiangensis

基　　金：国家自然科学基金(41371524);河南省基础与前沿研究(112300410283);河南省教育厅科学技术研究重点项目(13A630346);河南工程学院博士基金项目(D2013017)

摘　　要：利用统计年鉴中社会经济数据,选取DEA交叉模型对耕地利用效率进行评价,并引入Kohonen神经网络对评价结果进行聚类,分析了河南省耕地利用效率的空间分布特征。结果表明,DEA交叉效率评价模型可以避免出现伪有效决策单元,且效率评价值相对较低,说明评价结果客观、准确,避免了人为确定分类指标权重的主观性。投入水平对提高耕地利用效率有促进作用,而利用程度对耕地利用效率影响较小。河南省耕地利用效率较高且较为稳定的城市有南阳、商丘、濮阳、三门峡、许昌、新乡6个城市。聚类分析结果显示,河南省耕地利用效率的空间分布呈现明显区域差异,西北部黄河中下游平原地区耕地利用效率较高,而东部和南部地区则相对较低。In the present study,DEA intersection model was selected to evaluate cultivated land use efficiency based on social and economic data collected from the statistical yearbook,and the evaluation results were clustered with Kohonen neural network to analyze spatial distribution features of cultivated land use efficiency in Henan Province. It was shown that the DEA cross efficiency evaluation model could avoid spurious effective decision making units,and the evaluation value was relatively low,which indicated that the evaluation result was objective,accurate,and able to avoid the subjectivity of man-made classification index weight. The increasing input level could promote the utilization efficiency of farmland,while the influence of utilization degree on cultivated land use efficiency was minor. The cultivated land use efficiency in Nanyang,Shangqiu,Puyang,Sanmenxia,Xuchang,Xinxiang,was relatively high and stable. Cluster analysis exhibited obvious regional differences in spatial distribution of cultivated land use efficiency in Henan Province. The cultivated land use efficiency in the northwest plain area was high,while the cultivat-ed land use efficiency in the eastern and southern regions was relatively low.

关 键 词：耕地利用效率 DEA评价 交叉效率评价 KOHONEN神经网络 河南省 

分 类 号：S-9[农业科学] F301.2[经济管理—产业经济]