Closed dylanbeaudette closed 1 year ago
One such use case: ordering in sharpshootR::SoilTaxonomyDendrogram()
based on distance matrix from ordinal vs. nominal data.
I think this would be a good time to set up the package to build all of the datasets from a data-raw folder. I don't see any issues with replacing ST with this reordered version. It does not break tests, but would be a breaking change I think warranting a bump to 0.2.0
Here is some code for validating against other datasets in the package:
library(SoilTaxonomy)
load("./misc/ST-data/ST_logic.rda")
data("ST")
x <- data.frame(
order = getTaxonAtLevel(ST$subgroup, level = 'order'),
suborder = getTaxonAtLevel(ST$subgroup, level = 'suborder'),
greatgroup = getTaxonAtLevel(ST$subgroup, level = 'greatgroup'),
subgroup = ST$subgroup,
code = taxon_to_taxon_code(ST$subgroup)
)
x <- x[order(x$code),]
head(x)
#> order suborder greatgroup subgroup code
#> lithic folistels gelisols histels folistels lithic folistels AAAA
#> glacic folistels gelisols histels folistels glacic folistels AAAB
#> typic folistels gelisols histels folistels typic folistels AAAC
#> hemic glacistels gelisols histels glacistels hemic glacistels AABA
#> sapric glacistels gelisols histels glacistels sapric glacistels AABB
#> typic glacistels gelisols histels glacistels typic glacistels AABC
all(x$subgroup == ST_logic$subgroup)
#> [1] TRUE
Good call--and I agree 100%, we have to be able to re-generate all of this from canonical sources. The only thing I wasn't sure about re-generating had to do with creating the ST
object from NASIS domains. That is where it originally came from, but required a tiny but of manual "fixing". Lets sort this out at your convenience. A lot going on right now.
Closing this as the ordering was implemented in the ST_unique_list
object, not ST
See
misc/ST_logic.rda
, thanks to Marc Olds (UCD) for this suggestion. This would have an added benefit of setting reasonable factor levels based on the "order in which taxa key out" vs. alphabetical.