agrammonThe package agrammon facilitates using the Nitrogen flow model
Agrammon (www.agrammon.ch) via its REST API interface with the help of
R. The communication between the R session and the Agrammon model is
possible via Agrammon’s REST API (Wikipedia: REST
interface
and Agrammon REST
API.)
The package agrammon provides 6 functions that let the user
register_token(): store and re-use the personal access token
(section Access Token)save_template(): generate a valid data set (a CSV file) containing
all necessary input data (section Input Data
Set)run_agrammon(): run Agrammon via its REST interface (section Run
Agrammon)agrammon_defaults(): save some package specific defaults for the
current R session (section Run Agrammon)report(): extract report from a “full” set of model results
(section Run Agrammon)save_excel(): save results as an excel fileThe main function of this package run_agrammon() needs the path to a
CSV file with input data to run Agrammon. To help the user prepare a
valid input data file (i.e. a CSV file with all valid and necessary
inputs), the function save_template() will save a CSV template file
with all necessary input variables including their default values (if
the variable has a default) as well as a remark on the accepted input
values (if there are any restrictions). See section Input Data
Set for further explanation and examples on the
input data file.
run_agrammon() will act as follows:
data.frame or a data.table
(default)The user can process the results in the current R session or save them
as an EXCEL file using save_excel().
The package can be installed from GitHub. Use build_vignettes = TRUE
if you want this vignette available locally:
devtools::install_github('hafl-gel/agrammon-rest', build_vignettes = TRUE)To load & attach the package in the current R session, run:
library(agrammon)A personal access token (i.e., a string representing the user API key) is required to communicate with the Agrammon REST interface. This access token can be requested at support@agrammon.ch.
The access token can be registered either for the current R session or
permanently (in the .Renviron file in the user’s home directory
(Sys.getenv("HOME"))) using the function register_token(). If the
file .Renviron does not exist in the user home directory a new file
will be created, otherwise a line containing the token will be added to
the existing .Renviron file.
Most of the Agrammon input variables are mandatory and must contain
valid entries, only few input variables are optional and will fall back
on a predefined default. The function save_template() writes an input
template file that can be used as guidance. The template file column
separator is a semicolon (not a comma)!
Optional input variables contain an entry in the column ‘default’ in the template. Remarks on valid inputs (numeric bounds, list of valid entries, etc.) are indicated in column ‘remarks’ in the template.
A template file with all possible input variables (i.e. not limited to
specific animal categories) can be written by setting the argument
livestock to TRUE:
# all categories
save_template('template_all.csv', TRUE)
# template on all categories with german help text
save_template('template_all_de.csv', TRUE, language = 'de')The user can provide specific animal categories or their parent classes to restrict the template content to these categories:
# specific categories/classes
save_template('template_user.csv', list(RoughageConsuming = c('bisons', 'alpaca'),
wapiti = 'wapiti', dairy_cows = 'DC'), storage = 'tank 1')An example data file is shipped with the package.
# example input data set
path_ex <- system.file('extdata', 'example_data_set_3farms.csv', package = 'agrammon')
fread(path_ex)
#> V1 V2 V3 V4 V5
#> 1: ex_3farms 1 Livestock::OtherCattle[Beef Cattle]::Excretion dimensioning_barn 78
#> 2: ex_3farms 1 Application::Slurry::Csoft appl_hotdays sometimes
#> 3: ex_3farms 1 Application::Slurry::Cfermented fermented_slurry 0
#> 4: ex_3farms 1 Application::Slurry::Ctech share_deep_injection 0
#> 5: ex_3farms 1 Livestock::OtherCattle[Beef Cattle]::Housing::Floor mitigation_housing_floor none
#> ---
#> 303: ex_3farms 3 Application::Slurry::Cseason appl_autumn_winter_spring 50
#> 304: ex_3farms 3 Application::Slurry::Cseason appl_summer 50
#> 305: ex_3farms 3 Application::SolidManure::Cseason appl_autumn_winter_spring 50
#> 306: ex_3farms 3 Application::SolidManure::Cseason appl_summer 50
#> 307: ex_3farms 3 Application::Slurry::Csoft appl_evening 0The first column is optional and contains the simulation label. The second column is the farm label/id and is mandatory if the user specifies input for more than one farm.
The order of the columns is irrelevant since the package tries to find and assign the correct columns to the correct inputs. However, if you encounter issues that might be related to false column assignment, please report this back as an issue on GitHub (https://github.com/hafl-gel/agrammon-rest/issues).
With the path to the input data set file, the model can be run with the
run_agrammon() function.
Defaults to the function run_agrammon() can be saved on a R session
basis by using the function agrammon_defaults().
The content of the returned model result can be limited to contain
specific result reports. At the moment, the following reports can be
specified with argument report:
argument report |
model reporting |
|---|---|
"summary" |
Short summary on the “top-level” NH3 emissions |
"detailed" |
Summary on the “top-level” NH3 and other gaseous N emissions, as well as on the total N and TAN flow |
"NH3" |
Slightly more detailed summary on the NH3 emissions |
"TAN" |
Summary on the TAN flow |
"N" |
Summary on the total N flow |
"HAFL" |
Report used in teaching classes |
"full" |
Full (!) set of model results |
Argument report will be partially matched.
# path to example data set
path_ex <- system.file('inst/extdata', 'example_data_set_3farms.csv', package = 'agrammon')
# run model with default summary report:
run_agrammon(path_ex)
# run model with detailed results (report = 'detailed')
run_agrammon(path_ex, report = 'det')Instead of calling the Agrammon model with different reports several times, it is also possible to call Agrammon once with the “full” report returning and extract the reports afterwards. This allows to extract several reports by using only one model call which is faster, though, this means also an increased data traffic, since the full set of model results is returned (which might not be necessary if e.g. one is just interested in one specific report).
# path to example data set
path_ex <- system.file('inst/extdata', 'example_data_set_3farms.csv', package = 'agrammon')
# run model with 'full' report:
res <- run_agrammon(path_ex, report = 'full')
# extract N report
report(res, 'N')It is possible to “filter” the model results, i.e. attribute them to the different animal categories. The following filters can be specified:
argument filter |
animal category filtering |
|---|---|
"total_only" |
Only the total amount from all animal categories will be returned |
"existing_categories" |
Individual values incl. the total amount from all animal categories that exist in the input data set will be returned |
"all_categories" |
Individual values incl. the total amount from all animal categories that can be defined will be returned |
Argument filter will be partially matched.
# path to example data set
path_ex <- system.file('inst/extdata', 'example_data_set_3farms.csv', package = 'agrammon')
# default summary report incl. filter by all available categories (even if they don't exist
# in the input)
run_agrammon(path_ex, filter = 'all')
# detailed results incl. filter by existing categories
run_agrammon(path_ex, report = 'det', filter = 'ex')