标准化色标，面板中的 persp() 图，R

Question

我正在创建一个包含多个透视图的面板，每个图显示通过调整复杂回归模型中的预测变量生成的一系列数据。我希望我的 3D 表面由我的 z 变量进行颜色编码，我可以很容易地做到这一点。如何获取此色标并使其覆盖面板中所有图表示的所有 z 变量值的范围，以便可以比较多个透视图之间的色标？

#colorized z scale, only a smidge modified from http://www.inside-r.org/r-doc/graphics/persp
par(bg = "white")
par(mfrow=c(1,2))
x <- seq(-1.95, 1.95, length = 30)
y <- seq(-1.95, 1.95, length = 35)
z1 <- outer(x, y, function(a, b) a*b^2)
nrz <- nrow(z1)
ncz <- ncol(z1)
# Create a function interpolating colors in the range of specified colors
jet.colors <- colorRampPalette( c("blue", "green") )
# Generate the desired number of colors from this palette
nbcol <- 100
color <- jet.colors(nbcol)
# Compute the z-value at the facet centres
zfacet <- z1[-1, -1] + z1[-1, -ncz] + z1[-nrz, -1] + z1[-nrz, -ncz]
# Recode facet z-values into color indices
facetcol <- cut(zfacet, nbcol)
persp(x, y, z1, zlim=c(-8, 10.5), col = color[facetcol], phi = 30, theta = -30)
#zlim added to encompass combined ranges of z1 and z2

#exact same code, with different values of z
x <- seq(-1.95, 1.95, length = 30)
y <- seq(-1.95, 1.95, length = 35)
z2 <- outer(x, y, function(a, b) a*b^2+3)
nrz <- nrow(z2)
ncz <- ncol(z2)
# Create a function interpolating colors in the range of specified colors
jet.colors <- colorRampPalette( c("blue", "green") )
# Generate the desired number of colors from this palette
nbcol <- 100
color <- jet.colors(nbcol)
# Compute the z-value at the facet centres
zfacet <- z2[-1, -1] + z2[-1, -ncz] + z2[-nrz, -1] + z2[-nrz, -ncz]
# Recode facet z-values into color indices
facetcol <- cut(zfacet, nbcol)
persp(x, y, z2, zlim=c(-8, 10.5), col = color[facetcol], phi = 30, theta = -30)
#zlim added to encompass combined ranges of z1 and z2

#comparing the values, we see that there's different, overlapping scales from z1 and z2
summary(as.vector(z1))
summary(as.vector(z2))

因此，当您查看 z2 的两个图时，您会发现它们的比例不同。如果我能让色标覆盖 z1 和 z2 的范围，那就更清晰了

Answer 1

您可以在切割之前连接 z 值的两个向量（矩阵），然后在指定颜色时引用结果（切割）向量的相关子集。

color <- jet.colors(nbcol)
z1facet <- z1[-1, -1] + z1[-1, -ncz] + z1[-nrz, -1] + z1[-nrz, -ncz]
z2facet <- z2[-1, -1] + z2[-1, -ncz] + z2[-nrz, -1] + z2[-nrz, -ncz]
facetcol <- cut(c(z1facet, z2facet), nbcol)
persp(x, y, z1, zlim=c(-8, 10.5), 
      col=color[facetcol[seq_along(z1facet)]], 
      phi=30, theta=-30)
persp(x, y, z2, zlim=c(-8, 10.5), 
      col=color[facetcol[-seq_along(z1facet)]], 
      phi=30, theta=-30)

或者，使用lattice::wireframe，这可能更简单，特别是在处理许多曲面图时：

library(lattice)
# First, create a named list of the surface matrices.
L <- list(z1=z1, z2=z2)

# Iterate over the elements of L, creating data.frames with row number,
#  column number, and cell value, as well as matrix name (i.e. name of 
#  the element of L).
dat <- do.call(rbind, lapply(seq_along(L), function(i) {
  setNames(data.frame(which(!is.na(L[[i]]), arr.ind=TRUE),
             c(L[[i]]), names(L)[i]),
           c('x', 'y', 'z', 'g'))
}))

# Plot with wireframe, faceting by `g` (the matrix name).
wireframe(z ~ x + y|g, data=dat, drape=TRUE, col.regions=color)

根据需要进行调整screen以distance获得您想要的外观，例如：

wireframe(z ~ x + y|g, data=dat, drape=TRUE, col.regions=color, 
          screen=list(z=30, x=-60), distance=0.5)