Statistics 695V: Data Visualization, Spring 2005
Source: Columbia University
Many approaches to learning from data today involve the use of complex tools that extent tailor themselves to the patterns of the data, a form of automated learning. But human guidance from data visualizations added to the algorithms can vastly increase their performance. Other approaches involve the building of complex statistical models that are used to describe relationships among variables and to carry out probabilistic inductive inference. Data visualization enables the building of models that follow the patterns of the data, resulting in valid inferences.