efyj

A tool to easily run evaluation, adjustment or optimization of DEXi file (Multi-Attribute Decision Making).

• Copyright © 2014-2021 INRA
• The software is released under the MIT license. See the COPYING file.

Requirements

On Unix (Linux, OS X)

• libexpat (≥ 2)
• cmake (≥ 3.10)
• gcc ≥ 7.0 or clang ≥ 7.0 (need c++17 support)
• python ≥ 3.0 (optional)
• R ≥ 4.0 (optional)

For recent Debian and Ubuntu derivatives:

apt-get install build-essential cmake libexpat1-dev g++ python r-base r-base-dev

On Windows

Only necessary if you want to use the command line interface and/or the python package.

• Visual Studio 2017 or newer
• cmake (≥ 3.10)
• vcpkg
• vcpkg expat
• python ≥ 3.0 (optional)
• R ≥ 4.0 (optional)
• Rtools ≥ 4.0 (optional)

git clone https://github.com/microsoft/vcpkg
.\vcpkg\bootstrap-vcpkg.bat
.\vcpkg\vcpkg integrate install
.\vcpkg\vcpkg install

Python package installation

First, on Windows, make sure you have installed Visual Studio 2017, vcpkg and vcpkg expat. On Unix (Linux, OS X), install python and python-dev package.

Then, clone this repository and pip install. Note the --recursive option which is needed for both the pybind11 and fmt submodule:

git clone --recursive https://github.com/quesnel/efyj.git
pip install ./efyj

R package installation

On Windows, make sur you installed R 4.2.1, Rtools 4.2 or better. Then starts rtools bash from the start menu and type the following commands:

pacman -Sy
pacman -S git
pacman -S mingw-w64-i686-expat
pacman -S mingw-w64-x86_64-expat

On Unix (Linux and OS X), install R and R development files from the package tools of your distribution (apt, packman, yaourt, portage etc.).

Then, for both Windows and Unix, the first time you install efyj, clone this repository and install the package. Note the --recursive option needed for both the pybind11 and fmtlib submodules:

cd /c/users/XXXXXX/             # move to the right directory path
git clone --recursive https://github.com/quesnel/efyj.git

If you have already cloned efyj, you just need to update your copy. Use the following commands:

cd /c/users/XXXXXX/             # move to the right directory path
git fetch origin
git submodule update --init     # update modules if necessary
git reset --hard origin/master  # update source to master branch
git clean -dqxf                 # cleanup directory tree

Under a R terminal or Rstudio, type the following command (adapt the setwd command to the correct efyj clone):

install.packages("Rcpp")
library(Rcpp)
detach("package:Rcpp", unload = TRUE)
install.packages("devtools")
library(devtools)			# accept installation from source if required
install.packages("waldo")
setwd("C:/Users/XXXXXXXX/efyj/refyj")	# use the right directory path
library(roxygen2)
roxygen2::roxygenise()
load_all(".")
devtools::test()

Command line interface

Finally, efyj can be run in Windows or Unix terminal.

efyj -m file.dxi -o file.csv [-p] [-a]

The file.csv contains the options of the model. Four first one columns contains fields: site, fields, department, year. Next columns are basic attributes. Finally, the last column contains the result of the model dexi.

For example, if a DEXi file test.dxi have one aggregate attribute root and three basic attributes a, b and c, the CSV file is:

$ cat file.csv
site;fields;department;year;a;b;c;observed
xxxx;yyyy;1532;2010;ok;favourable;host;0-5
...

The '-p' option starts a prediction algorithm, '-a' option starts am adjustment algorithm.