landscapetools 0.5 about landscapetools

[marcosci]

Down to:

Imports: 
    classInt,
    ggplot2,
    raster,
    tibble

[mhesselbarth]

Currently, basically all funcitions have the prefix util_ no matter if they are e.g. for visualization or to convert data, right? I think it would make sense, to come up with a bit more detailed nameing scheme similar to landscapemetrics.

[marcosci]

Perfect, will do that :)
show_landscapes is roughly done, needs a bit of tweaking still.

[Nowosad]

Impressive!

[marcosci]

classInt can be taken of the list by implementing:

https://github.com/macroevolution/bammtools/blob/master/BAMMtools/src/jenksBrks.c

[Nowosad]

Talking about the landscapetools, is that expected behavior?

library(landscapetools)
l1 <- NLMR::nlm_fbm(64, 64)
l2 <- NLMR::nlm_planargradient(64, 64)
l3 <- NLMR::nlm_randomrectangularcluster(ncol = 60, nrow = 60, minl = 5, maxl = 10)
l4 <- NLMR::nlm_random(64, 64)

lst1 <- list(layer1 = l1,
             layer2 = l2,
             layer3 = l3,
             layer4 = l4)
util_facetplot(lst1, div_scales = FALSE)

util_facetplot(lst1, div_scales = TRUE)

^{Created on 2018-12-07 by the reprex package (v0.2.1)}

[marcosci]

Sure, do we want to keep show as prefix for the visualisation functions? Any ideas for merge, classify, rescale? 🤔 I think util could stay for raster2tibble for example, or?

And Jakub this was pretty much WIP 😆 I think util_facetplot will either way end up in util_plot as method and the idea with div_scales was to have a seperate color scale for each facet - to show spatial variables like pH, land use, ... at the moment the coloring works fine, but cowplot has to be replaced with grid.

[mhesselbarth]

For util_classify() I would suggest something as get_classes() or get_discrete() if we want to somehow stick to the naming style of landscapemetrics. As said before, all plot functions would get the show_-prefix. For the rest, I actually like the util_-prefix (e.g. util_merge() and util_rescale()).

[marcosci]

Just found this, may be helpful for show_landscapes: https://github.com/clauswilke/relayer

[marcosci]

util_writeESRI needs to be a bit faster, have to check how writeRaster makes that in raster.

[marcosci]

So, both of you now also have write access here.

I condensed it down to 6 dependencies, fonts are gone etc. and it should be ready for the merge with landscapemetrics 👍

[mhesselbarth]

Wait... did we decide to merge the two? Or do you mean to move some functions from landscapemetrics here?

[marcosci]

Move the get functions and vis stuff, or? 😅

…

On Mon, 21 Jan 2019, 23:13 Maximilian Hesselbarth ***@***.*** wrote: Wait... did we decide to merge the two? Or do you mean to move some functions from landscapemetrics here? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#14 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AKXHvhGJoErbtPHXhvJc98S2rp4IEcuWks5vFp4KgaJpZM4ZBFVy> .

[Nowosad]

@marcosci Would it be possible (and desired) to remove one or two more dependencies?

# devtools::install_github("jimhester/itdepends")
library(dplyr)
library(itdepends)

lst_dep = dep_usage_pkg("landscapetools")

lst_dep %>%
    group_by(pkg) %>% 
    count(sort = TRUE)
#> # A tibble: 9 x 2
#> # Groups:   pkg [9]
#>   pkg                n
#>   <chr>          <int>
#> 1 base             408
#> 2 ggplot2          154
#> 3 raster            40
#> 4 landscapetools    15
#> 5 grid               4
#> 6 dplyr              3
#> 7 tibble             3
#> 8 classInt           1
#> 9 tidyr              1

^{Created on 2019-01-22 by the reprex package (v0.2.1)}

[marcosci]

You mean dplyr. tibble or tidyr?
Sure, I can try that :-)
If we find an c++ implementation of the fisher algorithm, we could also get rid of classint.

[marcosci]

all done, if there is anything you would like to see before the merger let me know!