Bivariate kernel density graphs over a grid
tddens [, bwidth(proportion) sgraph bgraph sgopt(string) bgopt(string) gridpoints(integer) format(format) msize(markersizestyle) msymbol(symbolstyle) genstub(stubname) ]
tddens estimates bivariate density using a symmetric triangle kernel with bandwidth given as a proportion of sample range (option bwidth, with default 0.2) and produces several graphs. A "heat map" is produced by default; option sgraph shows a surface plot as points, and option bgraph shows a bar graph that offers an alternative view of the surface. The options sgopt and bgopt allow passing strings of options to the surface plot and bar graph, respectively. The options msize and msymbol apply to the surface plot only. Option gridpoints specifies how finely the grid of density estimates should be drawn; the default is 40 intervals in each dimension or 40+1 squared (1681) density estimates. Option format specifies formats for axis labels. Option genstub specifies a stub name for variable names to be saved on the current dataset, containing kernel density estimates and auxiliary variables used in the graphs.
Example
sysuse auto, clear tddens price mpg, s b tddens price mpg, ge(dens) Author of tddens
Austin Nichols <austinnichols@gmail.com>
Also see
On-line: help for spkde (if installed; findit spkde to download).