Mathematical Sciences: Applied Multivariate Tools

数学科学：应用多元工具

• 批准号: 8718362
• 负责人: Leo Breiman
• 金额: $ 13.78万
• 依托单位: University of California-Berkeley
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1988
• 资助国家: 美国
• 起止时间: 1988-05-15 至 1991-10-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=8718362&HistoricalAwards=false
• 关键词: Mathematical; Sciences; Applied; Multivariate; Tools

This research encompasses five topics aimed at the development of better working tools for the analysis of multivariate data: (1) Procedures for subset selection and evaluation in regression, (2) A new and better procedure for fitting nonlinear-nonparametric additive models through transformations of the predicted and the predictor variables, (3) Global nonlinear optimization methods, (4) Prototype analysis algorithms, and (5) Nonlinear smoothing methods. This research is in the statistical area of regression. Data analytic tools are extended following a significant step beyond traditional general linear models that was made practically feasible with the availability of cheap computational horsepower.

这项研究包括五个主题，旨在 开发更好的工作工具， 多变量数据：（1）子集选择和 （2）提出了一种新的、更好的方法， 拟合非线性-非参数加性模型， 预测变量和预测变量的变换，（3） 全局非线性优化方法;（4）原型分析 算法;（5）非线性平滑方法。 本研究属于回归统计领域。 数据分析工具在迈出重要一步后得到扩展 超越了传统的一般线性模型， 实际上可行的廉价计算的可用性 马力