{"title":"An Approach to Improve Diffusion Coefficient of Geospatial Information Model","authors":"Chongfu Huang","doi":"10.2991/dramclr-19.2019.1","DOIUrl":null,"url":null,"abstract":"This paper proposes an approach to improve the diffusion coefficient of the geospatial information diffusion model. The diffusion coefficient calculated by the average distance formula is appropriately amplified to become the initial diffusion coefficient. Employing a search method, we take two test points in the search interval consisting of 0 and the initial diffusion coefficient. Comparing the errors of the two test points used in the geospatial information diffusion model, we adjust the search interval: if the error of the left test point is small, the left point of the new search interval is unchanged, and the original right point of search interval is replaced with the right test point; if the error of the right test point is small, The right point of the new interval is unchanged, and the original left point of search interval is replaced with the left test point. Repeatedly, the search interval is continuously narrowed until the distance between the two test points is less than a given value, then the search is stopped. Meanwhile, the test point with a small error will be an optimized diffusion coefficient. A case constructing a relationship between the background data and disaster, with a sample size of 30, shows that the diffusion coefficient can reduce error approximately 17%. Keywords—geospatial information diffusion, diffusion coefficient, search interval, test point, background data, disaster 摘要—本文提出了一种改进地理空间信息扩散模型中扩散 系数的方法。将平均距离公式计算的扩散系数进行适当放 大,成为初始化扩散系数。使用某种搜索法,在由 0 和初始 化扩散系数构成的搜索区间中取两个测试点。将两个测试点 分别用于地理空间信息扩散模型,比较它们的误差,调整搜 索区间:如果左测试点的误差小,则新搜索区间左端点不 变,将原搜索区间的右端点换成右测试点;如果右测试点的 误差小,则新搜索区间的右端点不变,将原搜索区间左端点 换成左测试点。如此反复,不断缩小搜索区间,直到两个测 试点的距离小于一个给定的值,则停止搜索时,并以误差较 较的小测试点,为优化的扩散系数。用容量为 30 的样本,构 建背景数据和灾情间关系的算例表明,扩散系数优化后,大 约能减小 17%的估计误差。 关键词—地理空间信息扩散, 扩散系数, 搜索区间, 测试点, 背景数据 I. 引言 大灾中的信息孤岛,比比皆是。由于具有非线性识 别能力,且能学习矛盾样本,地理空间信息扩散模型 [1],较之加权地理回归[2]和人工神经元网络[3],能更好地 推测出空白地理单元上的灾情,有效解决信息孤岛的问 题。优化模型中的扩散系数,是进一步提高推测结果精 度的一个重要途径。 地理空间信息扩散模型,是将灾区已观测的地理单 元上的背景数据和灾情形成的样本,视为小样本,用正 态信息扩散公式[4],对其进行集值化处理,构造出“背 景数据”和“灾情”之间的因果关系。据此,我们用空 白地理单元上的背景数据,可推导出该地理单元上灾 情。 地理空间信息扩散模型,是一个集值统计回归模 型。扩散公式中的扩散系数,决定着样本点的集值化程 度,对预测结果有明显的影响。扩散系数较大时,较多 的监测点从一个样本点获得有效信息;反之,点较少。 理论上,样本足够大时,扩散系数为零,样本点的信 息,没有扩散。传统统计回归,就是在没有扩散的情况 下进行。换言之,传统统计回归,依赖于大样本。 目前,信息扩散理论的基础比较稳固,应用涉及面 较广。人们常用式(1)的平均距离公式计算正态信息 扩散中的扩散系数 h[5]。","PeriodicalId":142201,"journal":{"name":"Proceedings of the Fourth Symposium on Disaster Risk Analysis and Management in Chinese Littoral Regions (DRAMCLR 2019)","volume":"176 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the Fourth Symposium on Disaster Risk Analysis and Management in Chinese Littoral Regions (DRAMCLR 2019)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.2991/dramclr-19.2019.1","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
This paper proposes an approach to improve the diffusion coefficient of the geospatial information diffusion model. The diffusion coefficient calculated by the average distance formula is appropriately amplified to become the initial diffusion coefficient. Employing a search method, we take two test points in the search interval consisting of 0 and the initial diffusion coefficient. Comparing the errors of the two test points used in the geospatial information diffusion model, we adjust the search interval: if the error of the left test point is small, the left point of the new search interval is unchanged, and the original right point of search interval is replaced with the right test point; if the error of the right test point is small, The right point of the new interval is unchanged, and the original left point of search interval is replaced with the left test point. Repeatedly, the search interval is continuously narrowed until the distance between the two test points is less than a given value, then the search is stopped. Meanwhile, the test point with a small error will be an optimized diffusion coefficient. A case constructing a relationship between the background data and disaster, with a sample size of 30, shows that the diffusion coefficient can reduce error approximately 17%. Keywords—geospatial information diffusion, diffusion coefficient, search interval, test point, background data, disaster 摘要—本文提出了一种改进地理空间信息扩散模型中扩散 系数的方法。将平均距离公式计算的扩散系数进行适当放 大,成为初始化扩散系数。使用某种搜索法,在由 0 和初始 化扩散系数构成的搜索区间中取两个测试点。将两个测试点 分别用于地理空间信息扩散模型,比较它们的误差,调整搜 索区间:如果左测试点的误差小,则新搜索区间左端点不 变,将原搜索区间的右端点换成右测试点;如果右测试点的 误差小,则新搜索区间的右端点不变,将原搜索区间左端点 换成左测试点。如此反复,不断缩小搜索区间,直到两个测 试点的距离小于一个给定的值,则停止搜索时,并以误差较 较的小测试点,为优化的扩散系数。用容量为 30 的样本,构 建背景数据和灾情间关系的算例表明,扩散系数优化后,大 约能减小 17%的估计误差。 关键词—地理空间信息扩散, 扩散系数, 搜索区间, 测试点, 背景数据 I. 引言 大灾中的信息孤岛,比比皆是。由于具有非线性识 别能力,且能学习矛盾样本,地理空间信息扩散模型 [1],较之加权地理回归[2]和人工神经元网络[3],能更好地 推测出空白地理单元上的灾情,有效解决信息孤岛的问 题。优化模型中的扩散系数,是进一步提高推测结果精 度的一个重要途径。 地理空间信息扩散模型,是将灾区已观测的地理单 元上的背景数据和灾情形成的样本,视为小样本,用正 态信息扩散公式[4],对其进行集值化处理,构造出“背 景数据”和“灾情”之间的因果关系。据此,我们用空 白地理单元上的背景数据,可推导出该地理单元上灾 情。 地理空间信息扩散模型,是一个集值统计回归模 型。扩散公式中的扩散系数,决定着样本点的集值化程 度,对预测结果有明显的影响。扩散系数较大时,较多 的监测点从一个样本点获得有效信息;反之,点较少。 理论上,样本足够大时,扩散系数为零,样本点的信 息,没有扩散。传统统计回归,就是在没有扩散的情况 下进行。换言之,传统统计回归,依赖于大样本。 目前,信息扩散理论的基础比较稳固,应用涉及面 较广。人们常用式(1)的平均距离公式计算正态信息 扩散中的扩散系数 h[5]。