{"id":3665,"date":"2015-09-08T19:06:39","date_gmt":"2015-09-09T00:06:39","guid":{"rendered":"http:\/\/rud.is\/b\/?p=3665"},"modified":"2018-08-27T10:32:23","modified_gmt":"2018-08-27T15:32:23","slug":"roll-your-own-stats-and-geoms-in-ggplot2-part-1-splines","status":"publish","type":"post","link":"https:\/\/rud.is\/b\/2015\/09\/08\/roll-your-own-stats-and-geoms-in-ggplot2-part-1-splines\/","title":{"rendered":"Roll Your Own Stats and Geoms in ggplot2 (Part 1: Splines!)"},"content":{"rendered":"

A huge<\/em> change is coming to ggplot2<\/code> and you can get a preview of it over at Hadley’s github repo<\/a>. I’ve been keenly interested in this as I will be fixing, finishing & porting coord_proj<\/a> to it once it’s done.<\/p>\n

Hadley & Winston have re-built ggplot2 with an entirely new object-oriented system called ggproto<\/code>. With ggproto<\/code> it’s now possible to easily extend ggplot2 from within your own packages<\/em> (since source()<\/code> is so<\/em> last century), often times with very little effort.<\/p>\n

Before attempting to port coord_proj<\/code> I wanted to work through adding a Geom<\/code> and Stat<\/code> since thought it would be cool to be able to have interpolated line charts (and it helps answer some recurring StackOverflow “spline”\/ggplot2 questions) and also prefer KernSmooth::bkde<\/code> over the built-in density<\/code> function (which geom_density<\/code> and stat_density<\/code> both use).<\/p>\n

To that end, I’ve made a new github-installable package called ggalt<\/a> (h\/t to @jayjacobs for the better package name than I originally came up with) where I’ll be adding new Geom<\/code>s, Stat<\/code>s, Coord<\/code>s (et al) as I craft them. For now, let me introduce both geom_xspline()<\/code> and geom_bkde()<\/code> to show how easy it is to incorporate new functionality into ggplot2.<\/p>\n

While not a requirement, I think it’s a going to be a good idea to make both a paired Geom<\/code> and Stat<\/code> when adding those types of functionality to ggplot2. I found it easier to work with custom parameters this way and it also makes it feel<\/em> a bit more like the way ggplot2 itself works. For the interpolated line geom\/stat I used R’s graphics::xpsline<\/code> function. Here’s all it took to give ggplot2 lines some curves (you can find the commented version on github<\/a>):<\/p>\n

geom_xspline <- function(mapping = NULL, data = NULL, stat = \"xspline\",\n                      position = \"identity\", show.legend = NA,\n                      inherit.aes = TRUE, na.rm = TRUE,\n                      spline_shape=-0.25, open=TRUE, rep_ends=TRUE, ...) {\n  layer(\n    geom = GeomXspline,\n    mapping = mapping,\n    data = data,\n    stat = stat,\n    position = position,\n    show.legend = show.legend,\n    inherit.aes = inherit.aes,\n    params = list(spline_shape=spline_shape,\n                  open=open,\n                  rep_ends=rep_ends,\n                  ...)\n  )\n}\n\nGeomXspline <- ggproto(\"GeomXspline\", GeomLine,\n  required_aes = c(\"x\", \"y\"),\n  default_aes = aes(colour = \"black\", size = 0.5, linetype = 1, alpha = NA)\n)\n\nstat_xspline <- function(mapping = NULL, data = NULL, geom = \"line\",\n                     position = \"identity\", show.legend = NA, inherit.aes = TRUE,\n                     spline_shape=-0.25, open=TRUE, rep_ends=TRUE, ...) {\n  layer(\n    stat = StatXspline,\n    data = data,\n    mapping = mapping,\n    geom = geom,\n    position = position,\n    show.legend = show.legend,\n    inherit.aes = inherit.aes,\n    params = list(spline_shape=spline_shape,\n                  open=open,\n                  rep_ends=rep_ends,\n                  ...\n    )\n  )\n}\n\nStatXspline <- ggproto(\"StatXspline\", Stat,\n\n  required_aes = c(\"x\", \"y\"),\n\n  compute_group = function(self, data, scales, params,\n                           spline_shape=-0.25, open=TRUE, rep_ends=TRUE) {\n    tf <- tempfile(fileext=\".png\")\n    png(tf)\n    plot.new()\n    tmp <- xspline(data$x, data$y, spline_shape, open, rep_ends, draw=FALSE, NA, NA)\n    invisible(dev.off())\n    unlink(tf)\n\n    data.frame(x=tmp$x, y=tmp$y)\n  }\n)<\/code><\/pre>\n
If that seems like alot of code, it really isn't. What we have there are:<\/p>\n