Statistical Theory for Classification

分类统计理论

• 批准号: 9704327
• 负责人: John Hartigan
• 金额: $ 19万
• 依托单位: Yale University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1997
• 资助国家: 美国
• 起止时间: 1997-07-01 至 2000-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9704327&HistoricalAwards=false
• 关键词: Statistical; Theory; Classification

Hartigan 9704327 I explore the role of classification in probability and statistics. I use probability models in forming and validating classes, and in identifying modes and other new objects, but I also use classification for prediction and forming probability models. For example, in recognition problems, people compare new objects with classes of known objects, and predict behavior of these new objects by assigning the new objects to the classes and expecting the objects to behave similarly to known objects in those classes. I consider how properly formed classes may be used for understanding and prediction. For example, in congressional voting, members form a series of semi-permanent alliances or blocs that vote together on certain classes of issues. The size and constitution of these blocs offers a way to study the evolution of Congress.

Hartigan 9704327 我探讨了分类在概率和统计中的作用。我使用概率模型来形成和验证类，以及识别模式和其他新对象，但我也使用分类来预测和形成概率模型。例如，在识别问题中，人们将新对象与已知对象的类进行比较，并通过将新对象分配给类并期望这些对象的行为与这些类中的已知对象类似来预测这些新对象的行为。 我考虑如何使用正确形成的类来理解和预测。例如，在国会投票中，成员组成一系列半永久性联盟或集团，就某些类别的问题一起投票。这些集团的规模和构成为研究国会的演变提供了一种方法。