Adaptive Sampling

自适应采样

• 批准号: 9626102
• 负责人: Steven Thompson
• 金额: $ 20万
• 依托单位: Pennsylvania State Univ University Park
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1996
• 资助国家: 美国
• 起止时间: 1996-07-01 至 2001-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9626102&HistoricalAwards=false
• 关键词: Adaptive; Sampling

DMS 9626102 Thompson Adaptive sampling designs are designs in which the procedure for selecting the units to include in the sample may depend on observations made during the survey. For some populations, adaptive strategies produce substantial increases in precision of estimates compared to conventional designs with equivalent sample sizes. The research investigates new classes of adaptive designs and inference procedures and develops the needed basic theory. The research involves a combination of design-based and model-based approaches, and includes investigations in adaptive cluster sampling, adaptive allocation, adaptive graph sampling, optimal sampling strategies, multivariate methods in adaptive sampling including rapid-assessment strategies, and nonsampling errors in adaptive sampling. With adaptive sampling designs, the procedure for selecting the sample can be changed during a survey in response to observed patterns in the population. For example, in a survey of a rare, endangered animal species, whenever members of the species are detected, additional observations may be made at neighboring sites. In environmental pollution assessment studies, an adaptive plan allows additional observations to be made in the vicinity of observed "hot spots." Similarly, in surveys of hard-to-access human populations, social links between individuals may be adaptively used in obtaining the sample. Such designs are in marked contrast to conventional survey designs, in which the sites or people to be included in the sample can be determined prior to making any observations or obtaining any responses. Research in adaptive sampling has many important applications in environmental studies, including surveys of endangered species, natural resources, and environmental pollutants, as well as for scientific studies of hidden or hard-to-access human populations. Advantages of adaptive sampling strategies include the potential for obtaining better estimates of popu lation quantities with a given amount of sampling effort. Adaptive cluster sampling designs, for example, have been shown to be highly efficient relative to conventional designs for sampling populations that are very unevenly distributed. Adaptive designs can significantly increase the "yield" of the sample, so that, for example, more animals of the rare species are observed, while at the same time permitting unbiased estimation of quantities such as the total number of the animals in the population. Adaptive designs can be used to increase the probability of finding the values of most interest to investigators, such as the highest concentrations of a pollutant. In many scientific studies of hidden or hard-to-access human populations, adaptive link-tracing designs provide the only practical means for obtaining a sample large enough for study, so that effective methods for making estimates from such samples are vitally important. *** DMS 9626102 Thompson Adaptive sampling designs are designs in which the procedure for selecting the units to include in the sample may depend on observations made during the survey. For some populations, adaptive strategies produce substantial increases in precision of estimates compared to conventional designs with equivalent sample sizes. The research investigates new classes of adaptive designs and inference procedures and develops the needed basic theory. The research involves a combination of design-based and model-based approaches, and includes investigations in adaptive cluster sampling, adaptive allocation, adaptive graph sampling, optimal sampling strategies, multivariate methods in adaptive sampling including rapid-assessment strategies, and nonsampling errors in adaptive sampling. %%% With adaptive sampling designs, the procedure for selecting the sample can be changed during a survey in response to observed patterns in the population. For example, in a survey of a rare, endangered animal species, whenever members of the species are detected, additional observations may be made at neighboring sites. In environmental pollution assessment studies, an adaptive plan allows additional observations to be made in the vicinity of observed "hot spots." Similarly, in surveys of hard-to-access human populations, social links between individuals may be adaptively used in obtaining the sample. Such designs are in marked contrast to conventional survey designs, in which the sites or people to be included in the sample can be determined prior to making any observations or obtaining any responses. Research in adaptive sampling has many important applications in environmental studies, including surveys of endangered species, natural resources, and environmental pollutants, as well as for scientific studies of hidden or hard-to-access human populations. Advantages of adaptive sampling strategies include the potential for obtaining better estimates of population quantities with a given amount of sampling effort. Adaptive cluster sampling designs, for example, have been shown to be highly efficient relative to conventional designs for sampling populations that are very unevenly distributed. Adaptive designs can significantly increase the "yield" of the sample, so that, for example, more animals of the rare species are observed, while at the same time permitting unbiased estimation of quantities such as the total number of the animals in the population. Adaptive designs can be used to increase the probability of finding the values of most interest to investigators, such as the highest concentrations of a pollutant. In many scientific studies of hidden or hard-to-access human populations, adaptive link-tracing designs provide the only practical means for obtaining a sample large enough for study, so that effective methods for making estimates from such samples are vitally important. ***