Mathematical Sciences: Complexity and Information Based Criteria for Model Selection

数学科学：模型选择的复杂性和基于信息的标准

• 批准号: 9210131
• 负责人: Dominique Haughton
• 金额: $ 1.74万
• 依托单位: Bentley University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1992
• 资助国家: 美国
• 起止时间: 1992-07-15 至 1993-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9210131&HistoricalAwards=false
• 关键词: Mathematical; Sciences; Complexity; Information; Based

This research will deal with the properties of information as used in model selection. New information criteria have been developed which deal with situations of multiple data sets and/or with models which may have the same dimension. This research will expand on recent results for these criteria, and in particular will establish asymptotic properties. This research will deal with statistical model selection. The area, generally speaking, is concerned with the choice of the best model to fit a dat set among a set of candidate models. Issues of interest include problems of overfitting and the selection of variables in regression.

本研究将处理用于模型选择的信息属性。新的信息标准已经被开发出来，用于处理多个数据集和/或可能具有相同维度的模型的情况。本研究将扩展这些准则的最新结果，特别是将建立渐近性质。本研究将涉及统计模型的选择。一般来说，该领域关注的是在一组候选模型中选择最佳模型来拟合数据集。感兴趣的问题包括过拟合问题和回归中变量的选择。