srkobakian / sugarbag Goto Github PK

My centroids data had an x variable. This caused an issue downstream where lat_window was calculated as it was using the x in the centroids data instead of x in the environment (e.g. below with .env$x[1]).

lat_window <- function(x, cents = centroids, maximum = nlat) {
    max_int <- min(x[2], maximum)

    cents_in <- filter(cents, between(lat_int, .env$x[1], max_int))
    return(cents_in)
  }

Prepare for release:

• Check that description is informative
• Check licensing of included files
• usethis::use_cran_comments()
• devtools::check()
• devtools::check_win_devel()
• rhub::check_for_cran()
• Polish pkgdown reference index
• Draft blog post

Submit to CRAN:

• usethis::use_version('minor')
• Update cran-comments.md
• devtools::submit_cran()
• Approve email

Wait for CRAN...

• Accepted 🎉
• usethis::use_github_release()
• usethis::use_dev_version()
• usethis::use_news()
• Update install instructions in README
• Finish blog post
• Tweet
• Add link to blog post in pkgdown news menu

Can you take a look at these problems we see in revdep check for tidyr? The release of v1.0.0 is on the horizon. Looks like something related to name repair.

https://github.com/tidyverse/tidyr/blob/master/revdep/problems.md#sugarbag

The allocate() function assigns each geographical area to a hexagon on a pre-created grid.

allocate() has an argument focal_points. Users can supply a data frame of coordinates of map focal points, which are used to choose hex locations for each area that maximally preserve the relationship of the area to the focal point(s). A data frame capital_cities is provided containing lon/lat coordinates of Australian capital cities. This dataframe is used in the README example.

However, allocate() should work when focal_points = NULL, the default for the function. Line 85 in allocate.R contains code that should (I think) create a single focal point as the mean of all area centroids in the data where focal_points = NULL. But instead allocate() returns an error before getting to this point.

allocate() should work without focal point(s) being specified.

We need to be able to translate the polygon data from sf to tibble and back again. I recommend making a function to do this.

A lot of spatial data is now provided in UTM. Could you include sample code to show how to translate from UTM to lat/long for the hexmap creation?

If the wrong coordinate system is provided for the focal points an error message about out of memory

There seems to be two LNP in the readme:

auscolours <- c(
  "ALP" = "#DE3533",
  "LNP" = "#080CAB",
  "KAP" = "#b50204",
  "GRN" = "#10C25B",
  "XEN" = "#ff6300",
  "LNP" = "#0047AB",
  "IND" = "#307560")

while(NROW(f_grid) == 0) {
f_grid <- filter_grid_points(f_grid = hex_grid, f_centroid = centroid, f_dist = filter_dist)
if (NROW(f_grid) == 0) {
filter_dist <- filter_dist*1.5
print(paste("Filter Distance expanded 50% to ", filter_dist))
}
}

The endless loop occurs because there are no more available points in the direction, with the current buffer. You should be able to work out a maximum expansion size possible, based on the largest inter-grid point distance. When you reach this, break out of the while loop, print a warning to the user, and allocate to a randomly available grid point.

Expanding the buffer fixes the endless loop, temporarily.

"while"'s always need an escape route

Intended flow:
While no possible hexagon is available -> expand area considered to 1.5x original distance -> return points now available in the range

Current flow:
While no possible hexagon is available -> expand area considered to 1.5x original distance -> continues to expand, cannot break from while loop?

E.g. for create_grids, it wasn't obvious to me that the centroids argument required a data frame with variables explicitly named longitude and latitude. A flexibility to refer to any columns for lat and lon is desirable IMO. That way I don't need an intermediate step to create a data frame with latitude and longitude.

(happy to do pull request on this if you like)

For the data sets provided, add some code to show the data, and what would be done with the data. It might need to be wrapped in a don't run to satisfy CRAN

E.g. the use of as.tibble gives warning messages that it is deprecated.

Potential issues: may also require projection to be provided

when on downtime, needing something basic to do, write vignette

Develop a goodie bag of examples using the cancer data, to show off all sorts of aspects of the package

We want to be able to facet small chunks of the map, but the position in the facet should roughly match the geographic location. Like the geofacet package, but it only has US state layout currently. We would need to do a rough layout.