exercises/exo_dplyr.R

In nhenry50/datatoolboxexos2021: Code pour la formation datatoolbox

###############################################
# 29/11/2021, nicolas.henry@sb-roscoff.fr
#
# exo-dplyr.R
#
# script pour l'exo dplyr
#
###############################################

# load sp-eco data

dat_eco <- data_eco_list()
dat_sp <- data_sp_list()
dat_sp_eco <- data_sp_eco_list()
library(magrittr)

dat_eco |>
  dplyr::inner_join(dat_sp_eco,by = "ecoregion_id") |>
  dplyr::inner_join(dat_sp,by="species_id") |>
  dplyr::filter(family=="Ursidae"&sci_name!="Ursus malayanus") -> ursidae_table

ursidae_table |>
  dplyr::group_by(sci_name,species_id) |>
  dplyr::summarize(n_realm=dplyr::n_distinct(realm),
                   n_biome=dplyr::n_distinct(biome),
                   n_ecoregion=dplyr::n_distinct(ecoregion))

mam_per_eco <- table(dat$ecoregion_id)


dat_eco <- data_eco_list() |> data.table::data.table()
dat_sp <- data_sp_list() |> data.table::data.table()
dat_sp_eco <- data_sp_eco_list() |> data.table::data.table()


dat_merged <- dat_eco |>
  merge(dat_sp_eco,by="ecoregion_id") |>
  merge(dat_sp,by="species_id")

dat_merged[family=="Ursidae"&sci_name!="Ursus malayanus",
           lapply(.SD,function(X) length(unique(X))),
           by=list(species_id,sci_name),
           .SDcols=c("ecoregion","biome","realm")]