Brushing Scatterplots

TLDR

Dynamic graphical methods let analysts interact in real time with data displays, and brushing—selecting points or regions with a mouse—enables efficient exploration of multidimensional data, especially in scatterplot matrices. Brushing is implemented via a mouse‑controlled rectangle that can highlight, shadow‑highlight, delete, or label points, with transient, lasting, or undo paint modes and adjustable brush shapes, supporting techniques such as linking, conditioning, subsetting, and time‑series probing. An operation applied to one scatterplot instantly propagates to all other scatterplots in the matrix.

Abstract

A dynamic graphical method is one in which a data analyst interacts in real time with a data display on a computer graphics terminal. Using a screen input device such as a mouse, the analyst can specify, in a visual way, points or regions on the display and cause aspects of the display to change nearly instantaneously. Brushing is a collection of dynamic methods for viewing multidimensional data. It is very effective when used on a scatterplot matrix, a rectangular array of all pairwise scatterplots of the variables. Four brushing operations—highlight, shadow highlight, delete, and label—are carried out by moving a mouse-controlled rectangle, called the brush, over one of the scatterplots. The effect of an operation appears simultaneously on all scatterplots. Three paint modes—transient, lasting, and undo—and the ability to change the shape of the brush allow the analyst to specify collections of points on which the operations are carried out. Brushing can be used in various ways or on certain types of data; these usages are called brush techniques and include the following: single-point and cluster linking, conditioning on a single variable, conditioning on two variables, subsetting with categorical variables, and stationarity probing of a time series.