CODEWITHSUNDEEP
rsvginteractivelattice
user2808302
user2808302

Reputation:

Lattice Package in R: Is it possible to develop an interactive "scatterplot/network"?

I am developing an interactive scatterplot so that when the user rolls over a data point, a label is displayed. However, I would also like to add edges between certain data points.

I am successful at developing the interactive scatterplot using several libraries, including grid, gridSVG, lattice, and adegraphics. Below is a MWE:

library(grid)
library(gridSVG)
library(lattice)
library(adegraphics)

x = rnorm(10)
y = rnorm(10)
dat = data.frame(label = letters[1:10], x, y)

customPanel2 <- function(x, y, ...) {
  for (j in 1:nrow(dat)) {
    grid.circle(x[j], y[j], r = unit(.5, "mm"),
                default.unit = "native",
                name = paste("point", j, sep = "."))
  }
}

xyplot(y ~ x, panel = customPanel2, xlab = "x variable", ylab=NULL, scales=list(tck = c(1,0), y=list(at=NULL)))

for (i in 1:nrow(dat)) {
  grid.text(as.character(dat$label)[i], x = 0.1, y = 0.01, just = c("left", "bottom"), name = paste("label", i, sep = "."), gp = gpar(fontface = "bold.italic"))
}

for (i in 1:nrow(dat)) {
  grid.garnish(paste("point", i, sep = "."), onmouseover = paste('highlight("', i, '.1.1")', sep = ""), onmouseout = paste('dim("', i, '.1.1")', sep = ""))
  grid.garnish(paste("label", i, sep = "."), visibility = "hidden")
}

grid.script(filename = "aqm.js", inline = TRUE)
grid.export("interactiveScat.svg")

The resulting .svg file accomplishes everything I am aiming for - except that I also wish to add certain non-interactive edges. I tried to do this by incorporating the adeg.panel.edges method from the adegraphics library after defining the edges and the coordinates to be mapped. So, basically my xplot(...) function from before is replaced with:

edges = matrix(c(1, 2, 3, 2, 4, 1, 3, 4), byrow = TRUE, ncol = 2)
coords <- matrix(c(x[1], y[1], x[2], y[2], x[3], y[3], x[4], y[4]), byrow = TRUE, ncol = 2)

xyplot(y ~ x, panel = function(customPanel2){adeg.panel.edges(edges, coords, lty = 1:4, cex = 5)}, xlab = "x variable", ylab=NULL, scales=list(tck = c(1,0), y=list(at=NULL)))

It seems that this simply erases the interactive scatterplot made from the original xyplot, and simply outputs the static edge and coordinate image.

I tried to follow the example as seen in (http://finzi.psych.upenn.edu/library/adegraphics/html/adeg.panel.nb.html). Specifically, this example:

edges <- matrix(c(1, 2, 3, 2, 4, 1, 3, 4), byrow = TRUE, ncol = 2)
coords <- matrix(c(0, 1, 1, 0, 0, -1, -1, 0), byrow = TRUE, ncol = 2)
xyplot(coords[,2] ~ coords[,1],
  panel = function(...){adeg.panel.edges(edges, coords, lty = 1:4, cex = 5)})

I am a bit at a loss as to how to troubleshoot this problem, especially as I am mimicking the example code. Any suggestions are greatly appreciated!

Upvotes: 6

Views: 358

Answers (1)

skyebend
skyebend

Reputation: 1109

If what you are trying to produce is a node-link diagram of a network an alternate solution is to coerce your data into a network object and use the ndtv package to generate svg/htmlwidget interactive plots for your network. The ndtv package is designed for dynamic networks, but will generate interactive plots for static nets as well. 

library(ndtv)
data(emon) # load a list of example networks
render.d3movie(emon[[5]])  # render network 5  in the browser

Much more detail is in the tutorial http://statnet.csde.washington.edu/workshops/SUNBELT/current/ndtv/ndtv-d3_vignette.html

However, this does not use grid/lattice graphics at all

Upvotes: 1

Related Questions