Mathematical Sciences: Resampling Methods Under Long Range Dependence

数学科学：长程依赖下的重采样方法

• 批准号: 9505124
• 负责人: Soumendra Lahiri
• 金额: $ 7.5万
• 依托单位: Iowa State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1995
• 资助国家: 美国
• 起止时间: 1995-07-01 至 1998-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9505124&HistoricalAwards=false
• 关键词: Mathematical; Sciences; Resampling; Methods; Under

Proposal: DMS 9505124 PI: Saumendra Lahiri Institution: Iowa State University Title: Resampling Methods Under Long Range Dependence ABSTRACT: The main thrust of the research is to develop suitable resampling schemes and to study their properties for data exhibiting very strong forms of dependence. It is well known that existing resampling schemes perform very poorly under such dependence structures. Two new resampling schemes for long-range dependent data are investigated. These schemes are a sampling-window method based on subsamples of the data and a bootstrap method based on finite Fourier transformations. Their effectiveness in different inference problems is examined. The phenomenon of long range dependence, such as the concentration of an air-pollutant in a city observed at closely located sites, occurs when the data are highly correlated. This sort of data has a high level of variability, even after averaging. The research involves developing suitable resampling schemes for long-range dependent data that will provide good statistical estimates. The research has direct applications in Hydrology, Geology, Economics, and Environmental problems.

建议：DMS 9505124 PI：Saumendra Lahiri Institution：Iowa State University标题：长期相关性下的重采样方法摘要：研究的主要目的是开发合适的重采样方案，并研究其对表现出非常强相关性的数据的性质。众所周知，现有的重采样方案在这种依赖结构下的性能非常差。研究了两种新的长程相依数据重采样方案。这些方案是基于数据子样本的采样窗方法和基于有限傅里叶变换的Bootstrap方法。检验了它们在不同推理问题中的有效性。当数据高度相关时，就会出现长期相关性的现象，例如在靠近地点的地点观测到的城市空气污染物浓度。这类数据具有很高的可变性，即使在平均之后也是如此。这项研究涉及为长期相关数据开发适当的重新抽样方案，以提供良好的统计估计。这项研究在水文学、地质学、经济学和环境问题上都有直接的应用。