摘要
通过对聚类算法初始点选择策略的分析和比较,经典k-means算法在GIS海量数据处理上的效率问题,提出了随机采样的k-means算法来进行坐标聚类;并将随机采样k-means算法应用于GIS中心选址,充分利用GIS数据分析和处理能力,以城市间的欧几里得距离为相似条件,采用最大最小原则选取初始点进行聚类,从而缓解局部最优解产生的概率;选取中心城市作为目标对象,从而提高商业决策的充分性和可靠性;经仿真结果验证了所提出的随机取样k-means算法的有效性和正确率。
The limitation of classical k - means method is addressed in dealing with massive GIS data set through analyzing the several initialization strategies of algorithm. So a Sampling - Randomly k - means algorithm is presented to solve the clustering of GIS spatial data. Furthermore, the proposed algorithm is used to study the problem of the GIS centre location, and make decision - making for business more reliable. It can classify the world＇ s main cities and select the centroid city based on the Euclidean distance between two cities. And using the MaxMin Algorithm to select the initial points of k - means can reduce the probability of local optimization. Finally, the simulation results are given to demonstrate the effectiveness and correctness of the proposed algorithm.
出处
《计算机仿真》
CSCD
北大核心
2009年第9期256-260,共5页
Computer Simulation
基金
国家杰出青年科学基金资助(60525304)
浙江省科技攻关重点项目(2008C23040)
