In dis-organization/dpp: Projection with dplyr

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "README-"
)

dpp

dplyr for paired calculations

dpp depends on dplyr and imports proj4, and can be installed with

devtools::install_github("mdsumner/dpp")

In the function simd I apply a geographic transformation on two columns x,y in a tbl_df. To do this I extract the columns as a matrix and pass them to proj4::project, then pull them back to copy over the original longitude/latitude x,y columns.

The function is used like this:

library(dpp) 
x <- simd(n = 10, p = "+proj=laea")

and it works like this:

  x <- data_frame(x = runif(n, -180, 180), y = runif(n, -90, 90), id = seq(n))
  if (!is.null(p)) {
    m <- as.matrix(x[,c("x", "y")])
    m <- project(m, p)
    x$x <- m[,1L]
    x$y <- m[,2L]
  }

The question: is there a more dplyr-ish way to apply this paired variable mutation, without extracting to the matrix? Ultimately I'd like to have project methods for objects composed of tbl_dfs, so perhaps I need to write a new version of proj4::project that works more deeply with dplyr?

dis-organization/dpp documentation built on May 15, 2019, 8:51 a.m.