Wegman Edward J
Center for Computational Statistics, George Mason University, Fairfax, VA 22030-4444, USA.
Stat Med. 2003 May 15;22(9):1383-97. doi: 10.1002/sim.1502.
Data mining strategies are usually applied to opportunistically collected data and frequently focus on the discovery of structure such as clusters, bumps, trends, periodicities, associations and correlations, quantization and granularity, and other structures for which a visual data analysis is very appropriate and quite likely to yield insight. However, data mining strategies are often applied to massive data sets where visualization may not be very successful because of the limits of both screen resolution, human visual system resolution as well as the limits of available computational resources. In this paper I suggest some strategies for overcoming such limitations and illustrate visual data mining with some examples of successful attacks on high-dimensional and large data sets.
数据挖掘策略通常应用于机会性收集的数据，并且经常专注于发现诸如聚类、峰值、趋势、周期性、关联和相关性、量化和粒度等结构，以及其他视觉数据分析非常适用且很可能产生见解的结构。然而，数据挖掘策略经常应用于海量数据集，在这些数据集中，由于屏幕分辨率、人类视觉系统分辨率的限制以及可用计算资源的限制，可视化可能不太成功。在本文中，我提出了一些克服此类限制的策略，并通过对高维和大数据集成功进行分析的一些示例来说明可视化数据挖掘。
