Graphical data modeling methods using ceres plots

Identification of curvature in regression models is an important aspect of data analysis. Partial residual plots have played a major role. Recently, Cook developed a new class of plots, called CERES plots, that includes partial residual plots. Implementation of these plots necessitates modeling the relationships between certain covariates. If these relationships are linear, a partial residual plot is obtained. However, if the relationships are nonlinear, the more general CERES plot is obtained. Here, we describe an implementation of interactive methods which construct CERES plots.