Here you can access the source scripts of the geoprocessing tools built upon the arcgisbinding package. The three tools down below are thought to be executed sequentially and will help you to get up and running with the Bridge. This toolset if focused on data retrieval from the Eurostat repository with the eurostat package, data processing and mapping using the tricolore package, iplemented and executed in the ArcGIS desktop software.
R-ArcGIS bridge, developed and maintained by Esri, is a little-known secret to extend the functionality of geoprocessing tools in ArcGIS.
As a data scientist, you might be already using several tools that help you find answers to sometimes very complicated questions about the patterns and relationships ocurring in your data. R is one of those tools that provides a variety of packages containing functions for (geo)statistical analysis. Using the bridge between R and ArcGIS, you can easily access geographic data and take advantage of ArcGIS software while performing your analysis using R on the background. It is also possible to generelize your work and convert these R scripts into geoprocessing tools so you or anyone you want to share it with could run them directly in ArcGIS as a typical standalone or built-in in a (e.g. ModelBuilder) workflow tool.
Get started with the R-ArcGIS Bridge :
The full code for this tool is provided in the eurostat_download_data_tool.R script.
Main steps this script goes through:
👉 Definition of the input and output parameters of the script, in the exact order it will be specified by the user in the geoprocessing tool interface:
input_toc_table <- in_params[[1]] # csv table provided to you
input_toc_title <- in_params[[2]] # title of user choice
input_nuts_spain_limits <- in_params[[3]] # limit or not to Spain, possible options "Yes" or "No"
input_nuts_level <- in_params[[4]] # level of dissagregation, possible options 0,1,2 or 3
output_nuts <- out_params[[1]] # where the output spatial data will be saved to
output_dic <- out_params[[2]] # label dictionary, non-spatial datasetPlease note that eurostat_toc.csv was provided for your convenience, but you can always generate the last update by running the following:
get_eurostat_toc() %>%
filter(type %in% c("dataset", "table")) %>%
distinct() %>%
write.csv(file = "eurostat_toc.csv")👉 Retrieval of the spatial data for selected NUTS level. If you want to limit your data to Spain you can set up a conditinal statement e.g.:
if(input_nuts_spain_limits == "Yes"){
nuts <- get_eurostat_geospatial(output_class = "sf",
resolution = "60",
nuts_level = input_nuts_level) %>%
select(-c(id, NUTS_NAME, FID)) %>%
filter(CNTR_CODE == "ES")
...
}👉 Getting the code for the selected table from the pre-loaded list:
arc.open(input_toc_table) %>%
arc.select(where_clause = input_toc_title) %>%
pull(code) -> cc👉 Direct data download from the source repository:
as.data.frame(get_eurostat(id = cc))%>%
label_eurostat(fix_duplicated=T, code = "geo") %>%
mutate_if(is.factor, as.character) %>%
as.data.frame() %>%
mutate(time = as.POSIXct(time))-> data👉 Merging retrieved dataset with the spatial data and writing it to the output location provided by the user:
arc.write(output_nuts, merge(nuts, data, by = "geo_code"))When the script is completed and wrapped in tool_exec<- function(in_params, out_params){...} you should create a new script in an ArcGIS toolbox, either new or pre-existing. In the script properties link it to the R code location on the disk:
Once it´s done, move to the Parameters and specify each of the input and output variables in the order you list it in the script. You should come up with 4 input ⬇️ (Table, SQL Expression, Character and Numeric) and 2 output ⬆️ (Feature Class and Data Table) variables:
The inteface of this brand new tool looks like a traditional geoprocesing ArcGIS tool, however it uses your R code on the background.
data_transformation_tool.R corresponds to the second step in the process of preparing your data so it´s suitable for ternary map generation.
👉 Following the same logic explained in the Step 1, let´s set up inputs and outputs:
input_data <- in_params[[1]] # spatial data you created at the previous step
input_field_to_transpose <- in_params[[2]] # name of the variable you want your ternary composition for
input_filter_expression <- in_params[[3]] # filtering expression for your dataset to avoid ambiguity
output_data <- out_params[[1]] # transformed spatial data
output_dic <- out_params[[2]] # labelling dictionary👉 Now proceed with some simple data wrangling:
arc.open(input_data) %>%
arc.select(where_clause = input_filter_expression) %>%
arc.data2sf() %>%
select(geo, input_field_to_transpose, values) %>%
set_colnames(c("geo", "variable", "value", "geometry")) %>%
left_join(get_eurostat_dic(input_field_to_transpose, lang = "en") %>%
as.data.frame() %>%
set_colnames(c("code", "variable")),
by = "variable") %>%
select(-variable) %>%
dcast(geo ~ code) -> d_transThe object obtained on the previous step does not belong to a spatial class and in order to return a feature class to the location specified by user in
out_params[[1]]we will need to add the spatial attributes back to the data framed_trans.
One you have successfully completed the R code and created a new script in the same toolbox in ArcGIS (linking your tool to the source code in the General page), you may proceed to the next step and specify the Parameters:
Remember that you always need to respect the order of the input and output parameters. The final outlook of your tool might look like the one below:
You are now ready to create you ternary composition map. Here we will be making use of a great tricolore package developed by Jonas Schöley and Ilya Kashnitsky. For more information on this package please refer to the vignette.
👉 You will start by specifying ⬇️ inputs (entry feature class input_data <- in_params[[1]] and 3 variables (input_var1 <- in_params[[2]]; input_var2 <- in_params[[3]]; input_var3 <- in_params[[4]]) you want you ternary map to be based on) and ⬆️ outpus (in this case, you will create one feature class output_data_vector <- out_params[[1]] with several fields in the attribute table containing information on the color codes in different formats, 3 rasters, one for red output_data_raster_red <- out_params[[2]], green output_data_raster_green <- out_params[[3]] and blue output_data_raster_blue <- out_params[[4]] bands, and one static pdf report output_report <- out_params[[5]]).
👉 Once the data is successfully read into ArcGIS and set to a spatial dataframe with d <- arc.open(input_data) %>% arc.select(fields = fields_list) %>% arc.data2sf() you are ready to run the Tricolore function in order to color-code your data and generate a color key:
# Color-code three-part compositions with a ternary balance color scale and return a color key
tric_var <- Tricolore(d, p1 = input_var1, p2 = input_var2, p3 = input_var3)
# Adding back the vector of colors to the spatial data frame
d$hex <- tric_var$rgb
# converting hex code to rgb values
cbind(d, t(col2rgb(d$hex))) -> d
# adding an rgb formula into as a separate field to the dataframe
d$rgb <- paste0("rgb(", d$red, ", ", d$green, ", ", d$blue, ")")Remember that you can always customize the messages that appear in the pop-up window with
arc.progress_label()along with the bar progress labelarc.progress_pos(). If you would also like some text output to be displayed or graphics generated while tool is executing you can simply force it to be visualized with theprint()command.
👉 arcgisbinding package supports both raster and vector formats. For the demonstration purposes you will rasterise the vector dataset that you created on the previous step following the same sequence for all RGB values, e.g.:
r_red <- fasterize(d, raster(d, res = 2000), field = "red", fun="sum")👉 On the last step you will generate a pdf report containing:
- 📍 a static map:
ggplot(d) + geom_sf(aes(fill = hex, geometry = geom), size = 0.1) + scale_fill_identity() - 🎨 a color legend:
tric_var$key - 📊 plots with group comparison in between-subjects designs created with the ggstatsplot package:
ggbetweenstats(data = dd, x = variable,y = value)
The Parameters of this final script need to be specified as follows:
As mentioned earlier, please check the numbering of the in_params[[]] and out_params[[]] of your script against the order they eventually enter your tool:
To access the full code please go to the tricolore_mapping_tool.R.
One of the most powerful sides that R-ArcGIS offers is the possibility to include your tools built upon the R code into the traditional geoprocessing workflow.
The model above goes throw the three steps of downloading, transfroming and mapping the data and eventually combines three output rasters of red, green and blue bands into a singular composite unit. The legend is generated on the fly and displayed to user in a pop-up window.
If you´re experiencing any troubles with parameter specification feel free to download this ready-to-use toolbox 🎁