This dialog sets up a call to the scatter3dHH function to draw a
three-dimensional scatterplot, and optionally to Identify3d to label
points interactively with the mouse.
John Fox jfox@mcmaster.ca. Squared residuals added by Richard M. Heiberger <rmh@temple.edu>.
The explanatory variables provide the "horizontal" and "out-of-screen" axes of the scatterplot, the response variable provides the "vertical" axis.
Data points are represented as spheres or points, depending upon the number of observations.
Several regression surfaces can be
plotted: a linear least-squares surface; a full quadratic least-squares surface
with squared and cross-product terms; a "smooth" regression surface --- either a
smoothing spline, if no degrees of freedom are specified (in which case the
gam function selects the df by generalized cross validation),
or a fixed-df regression spline; an additive-regression surface (also fit by gam),
with either smoothing spline or regression spline components (again selected according
to the specification of degrees of freedom). If only one surface is fit, then residuals are
plotted as red (negative) and green (positive) lines from the surface to
the points. If the squared residuals option is checked, then squared
residuals are plotted. The sum of the area of these squares is the
"residual sum of squares".
You can specify a factor defining groups by pressing the Plot by groups button. A separate surface or set of surfaces is plotted for each level of the groups factor. These surfaces can be constrained to be parallel.
The completed plot can be manipulated with the mouse: Click, hold, drag the left mouse button to rotate the display; click, hold, and drag the right button (or centre button on a three-button mouse) to zoom in and out.
If the box labelled Identify observations with mouse is checked, you may use the mouse to identify points interactively: Press the right mouse button (or the centre button on a three-button mouse), drag a rectangle around the points to be identified, and release the button. Repeat this procedure for each point or set of "nearby" points to be identified. To exit from point-identification mode, right-click (or centre-click) in an empty region of the plot.
Points may also be identified subsequently by selecting Identify observations with mouse from the R Commander 3D graph menu: As above, click and drag the left mouse button to rotate the display, and click and drag the right (or centre) button to identify points.
scatter3dHH, Identify3d,
rgl.open, gam