panel.hexbin(x, y, subscripts, hexbin.data, lwd=add.line$lwd, lty=add.line$lty, col=add.line$col, ..., lmline=F, text=!fill, fill=T, maxcount=max(c(0, bin$count)), mincount=1, col.regions=trellis.par.get("regions")$col, count.to.col, border=F) panel.hexbin.lmline(x, y, subscripts, hexbin.data, lwd = add.line$lwd, lty = add.line$lty, col = add.line$col, ...) panel.hexbin.loess(x, y, subscripts, hexbin.data, lwd = add.line$lwd, lty = add.line$lty, col = add.line$col, ..., span = 2/3, degree = 1, family = c("symmetric", "gaussian"), evaluation = 50)
hexbin.data[subscripts,][,c("x","y")]
x
and
y
observations originate.
hexbin
, as produced by the
hexbin
function.
bin=hexbin.data[subscripts, , drop = F]
is the part of
hexbin.data
that belongs in the current panel.
The bin's
x
and
y
components provide the
"center of gravity" of the points in each cell (and are the
x
and
y
arguments to
panel.hexbin
).
cell2xy(bin)
provides the the geometric centers of the cells.
bin$count
gives the counts in each cell.
lmline
is set to
TRUE
.
lmline
is set to
TRUE
.
lmline
is set to
TRUE
.
TRUE
, print the count in each cell.
FALSE
, do not color in the hexagonal cells.
count.to.col
are subscripts into this list
of colors.
TRUE
, draws borders around each hexagonal cell.
panel.hexbin.loess
.
1
is locally-linear fitting and
2
is locally-quadratic fitting.
"gaussian"
or
"symmetric"
. In the first case,
local-fitting methods are used. In the second case, the default, local
fitting is used together with a robustness feature that guards against
distortion by outliers.
x
and
y
values.
The
x
and
y
components are required by the
underlying Trellis function, but for this function, they are supplied
by
hexbin.data
.
library(bigdata) xyplot(data=as.bdFrame(fuel.frame), Disp.~Weight|Type, panel = function(...) { panel.hexbin(...)}) #as.bdFrame requires loadin the big data library.