摘要
以四川省的地形、气候为研究对象,针对山地地形特征与气候变化研究中,传统的统计分析、非线性拟合等方法缺乏分析处理海量数据和提取隐含信息能力的问题,提出将关联规则数据挖掘与栅格图像处理、地形分析相结合的研究方法。该方法利用栅格图像处理和地形分析技术,对地形和气候栅格图像进行坐标转换、裁剪、分类、因子提取、离散化等预处理,再用Apriori算法对提取的地形特征因子和气候因子进行分析,得到反映两者之间相关性的强关联规则。通过对60余万组数据的分析,得到22条满足最小支持度和置信度的关联规则,并由此综合分析得到6条复合关联规则。实验证明,这些反映地形特征与气候变化幅度之间关联性的关联规则可信度较高。
In this paper, aimed at problems which commonly research means, such as statistical analysis, nonlinear fitting and so on, lack the ability of extraction hidden information and processing mass data in mountain terrain feature and climate change research, the terrain and climate of Sichuan is taken as research object, and a research method which combines association rules data mining, raster image processing and terrain feature factors extraction is put forward. In this method, raster image processing and terrain analysis technique are adopt to pre-treat terrain and climate raster image by coordinate transformation, clipping, classification,factors extraction,discretization and so on,and Apriori algorithm is used to analysis terrain features and climate factors extracted by above pretreatment methods for obtaining strong association rules which represent relativity law between terrain and climate change. By means of analysis on more than 600 000 group data, totally 22 items association rules which satisfy minimum support and confidence are obtained, and 6 compound association rules are generalized by comprehensive analysis as well. Experiment shows that those results have higher credibility in reflecting associated laws between terrain feature and climate change.
出处
《地理与地理信息科学》
CSSCI
CSCD
北大核心
2010年第1期37-40,47,共5页
Geography and Geo-Information Science
基金
国家自然科学基金(40972225)
国家科技支撑"十一五"计划(2008BAK49B02)
国家863重点项目(2007AA120306)
四川省杰出青年学科带头人培养计划项目(06ZQ026-014)
四川省教育厅自然科学重点项目(2006A116)
关键词
地形
山地
气候
关联规则
数据挖掘
terrain
mountain
climate
association rules
data mining