This repository contains the implementation of the algorithms and data of the experimental section of the paper "Aggregating Value Systems for Decision Support" by Roger X. Lera-Leri, Enrico Liscio, Filippo Bistaffa, Catholijn M. Jonker, Maite Lopez-Sanchez, Pradeep K. Murukannaiah, Juan A. Rodríguez-Aguilar, and Francisco Salas-Molina in Knowledge-Based Systems, 2024.

• Python 3
• Julia and PyJulia
• Pandas
• Csv library
• Numpy
• CVXPY
• ECOS (Embedded within CVXPY)

All experiments consider the European Values Study 2017: Integrated Dataset (EVS 2017) (dataset).

Our approach must be executed by means of the solve.py Python script, i.e.,

usage: solve.py [-h] [-p P] [-e E] [-f F] [-w W] [-i I] [-o O] [-v] [-l] [-t]
                [-g G]

optional arguments:
  -h, --help  show this help message and exit
  -p P        p-norm (default: 2)
  -e E        epsilon used to compute limit p (default: 1e-4)
  -f F        CSV file with data (default: 'data.csv')
  -w W        weighting countries: 0 for unweighted problem, 1 for considering people that participated in the study and 2 for country population (default: 0)
  -i I        computes equivalent p given an input consensus
  -o O        write consensus to file
  -v          computes the preference aggregation
  -l          compute the limit p
  -t          compute the threshold p
  -g G        store results in csv

This repository contains the implementation of the pIRLS algorithm (article). This article should be cited when citing our work.

• Install requirements: pip install -r requirements.txt
• To run locally with PyJulia compatability issues
  • Run python-jl -m pip install IPython to install IPython in Julia
  • Run your command as usual using python-jl -m IPython instead of python

As an example: python-jl -m IPython solve.py -f toy_data.csv -v true -g results.csv wil:

• Load the toy_data.csv file
• Compute the preference aggregation
• Store the results in results.csv