Beyond Control Panels: Direct Manipulation for Visual Analytics

To tackle the onset of big data, visual analytics seeks to marry the human intuition of visualization with mathematical models' analytical horsepower. A critical question is, how will humans interact with and steer these complex models? Initially, users applied direct manipulation to such models the same way they applied it to simpler visualizations in the premodel era--using control panels to directly manipulate model parameters. However, opportunities are arising for direct manipulation of the model outputs, where the users' thought processes take place, rather than the inputs. This article presents this new agenda for direct manipulation for visual analytics.