Data Mining, Statistical Learning, and Data Visualization for Complex Data

标题：复杂数据的数据挖掘、统计学习和数据可视化首席研究员：William S.尚蒂·克利夫兰普渡大学统计学教授和计算机科学礼貌教授这个项目研究了六个研究项目的计算环境的构建，这些研究项目通过对复杂数据建模的需要而结合在一起--具有复杂模式和依赖关系的大型数据集，需要复杂的模型和算法。 其中三个项目的目标是数据分析：使用数据挖掘，统计学习和数据可视化的方法，通过分析复杂的数据集来研究主题领域的问题。 其他三个目标是新方法：开发新的数据挖掘方法和工具，统计学习和复杂数据的数据可视化。要使方法最好地用于复杂数据，就必须考虑到复杂数据的挑战。分析复杂数据所需的环境以及为它们发明新方法与网格计算环境有很大的不同，网格计算环境在过去受到了很大的关注。数据分析需要（1）一个交互式环境，具有良好的数据编程语言，提供新方法的快速原型，以及（2）支持交互式计算的硬件环境，具有大量内存和对大型数据库的快速磁盘访问。 该项目的一个目标是研究如何设计这样一个环境，并使成本最小化。

Title: Data Mining, Statistical Learning, and Data Visualization forComplex DataPrincipal Investigator: William S. Cleveland, Shanti S. Gupta Professor ofStatistics and Courtesy Professor of Computer Science, Purdue UniversityThis project studies the building of a computational environment for sixresearch projects that are bound together by the need to model data thatare complex --- large data sets with complex patterns and dependenciesthat require require complex models and algorithms. Three of theprojects have as their goal data analysis: using methods of data mining,statistical learning, and data visualization to study the problems in asubject matter area through the analysis of complex data sets. The otherthree have as their goal new methods: the development of new approachesand tools of data mining, statistical learning, and data visualizationfor complex data. For methods to work best for complex data, they mustbe invented with the challenges of complex data in mind.The environment needed for the analysis of complex data and the inventionof new methods for them is quite different than that of grid computingenvironments, which have received substantial attention in the past. Dataanalysis requires (1) an interactive environment with a good language forprogramming with data that provides rapid prototyping of new methods,and (2) a hardware environment that supports interactive computing with largeamounts of memory and rapid disk access to large databases. A goal of theproject to study how to design such an environment and minimize cost.