Workshop/Conference for Probability Models and Statistical Analyses for Ranking Data

概率模型和排名数据统计分析研讨会/会议

• 批准号: 0428026
• 负责人: Joseph Verducci
• 金额: --
• 依托单位: Ohio State University Research Foundation -DO NOT USE
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-05-15 至 2005-04-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0428026&HistoricalAwards=false
• 关键词: Workshop; Conference; Probability; Models; Statistical

abstract PI: Joseph Verducciproposal: 0428026Workshop/Conference for Probability Models and Statistical Analyses for Ranking DataThe workshop, geared toward young researchers, is comprised of five week-long background courses and a lecture series of current research topics. The five short courses for ranking data are 1) Probability Models; 2) Experimental Designs and Analysis; 3) Simulation Based Methods; 4) Paired Comparisons; 5) Tests of Independence. Current research topics include Ranking Methods for Search Engines, Connections with the Theory of Group Representations, and Random Effects in Consumer Preferences. The conference, geared more toward specialists, consists of invited and contributed papers from diverse areas, including Machine Learning, Algebra, and Utility Theory. Contributed papers emphasize current applications.Data in the form of rankings may be found everywhere: in marketing, rankings are the key to studying subtle changes in consumer preferences under changing conditions; in sports, for determining everything from college team seedings to Olympic medals; on the internet, thousands of similar sites must be ranked in different ways to meet the needs of diverse clients. Accurate analysis of such data requires specialized methods. Historically these methods were studied in three eras, the pre-computer era, where simple models were constructed, largely based on theories from psychology and economics; the early computer era, where many models were categorized and implemented; and the modern era, where models are being tailored to complex problems. The workshop/conference is designed to lead young researchers through the diverse literature to the current state of the art.

该研讨会面向年轻的研究人员，由为期五周的背景课程和当前研究主题的系列讲座组成。对数据进行排序的五个短期课程是：1)概率模型；2)实验设计与分析；3)基于模拟的方法；4)配对比较；5)独立性测试。目前的研究主题包括搜索引擎的排名方法，与群表示理论的联系，以及消费者偏好的随机效应。这次会议更多地面向专家，由来自不同领域的受邀论文和投稿论文组成，包括机器学习、代数和效用理论。投稿论文强调当前的应用。排名形式的数据随处可见：在营销领域，排名是研究不断变化的条件下消费者偏好细微变化的关键；在体育领域，用于确定从大学团体种子选手到奥运奖牌的一切；在互联网上，数千个类似的网站必须以不同的方式进行排名，以满足不同客户的需求。对这些数据的准确分析需要专门的方法。从历史上看，这些方法是在三个时代进行研究的：前计算机时代，构建简单的模型，主要基于心理学和经济学的理论；早期计算机时代，许多模型被分类和实施；以及现代时代，模型正在为复杂的问题量身定做。研讨会/会议旨在引导年轻的研究人员通过不同的文献达到当前的艺术水平。