Multiple-criteria decision-making (MCDM) with Electre, Promethee, Weighted Sum and Pareto

Last update: Nov 22, 2022

Overview

EasyMCDM - Quick Installation methods

Install with PyPI

Once you have created your Python environment (Python 3.6+) you can simply type:

pip3 install EasyMCDM

Install with GitHub

Once you have created your Python environment (Python 3.6+) you can simply type:

git clone https://github.com/qanastek/EasyMCDM.git
cd EasyMCDM
pip3 install -r requirements.txt
pip3 install --editable .

Any modification made to the EasyMCDM package will be automatically interpreted as we installed it with the --editable flag.

Setup with Anaconda

conda create --name EasyMCDM python=3.6 -y
conda activate EasyMCDM

More information on managing environments with Anaconda can be found in the conda cheat sheet.

Try It

Data in tests/data/donnees.csv :

alfa_156,23817,201,8,39.6,6,378,31.2
audi_a4,25771,195,5.7,35.8,7,440,33
cit_xantia,25496,195,7.9,37,2,480,34

Promethee

from EasyMCDM.models.Promethee import Promethee

data = pd.read_csv('tests/data/donnees.csv', header=None).to_numpy()
# or
data = {
  "alfa_156": [23817.0, 201.0, 8.0, 39.6, 6.0, 378.0, 31.2],
  "audi_a4": [25771.0, 195.0, 5.7, 35.8, 7.0, 440.0, 33.0],
  "cit_xantia": [25496.0, 195.0, 7.9, 37.0, 2.0, 480.0, 34.0]
}
weights = [0.14,0.14,0.14,0.14,0.14,0.14,0.14]
prefs = ["min","max","min","min","min","max","min"]

p = Promethee(data=data, verbose=False)
res = p.solve(weights=weights, prefs=prefs)
print(res)

Output :

{
  'phi_negative': [('rnlt_safrane', 2.381), ('vw_passat', 2.9404), ('bmw_320d', 3.3603), ('saab_tid', 3.921), ('audi_a4', 4.34), ('cit_xantia', 4.48), ('rnlt_laguna', 5.04), ('alfa_156', 5.32), ('peugeot_406', 5.461), ('cit_xsara', 5.741)],
  'phi_positive': [('rnlt_safrane', 6.301), ('vw_passat', 5.462), ('bmw_320d', 5.18), ('saab_tid', 4.76), ('audi_a4', 4.0605), ('cit_xantia', 3.921), ('rnlt_laguna', 3.6406), ('alfa_156', 3.501), ('peugeot_406', 3.08), ('cit_xsara', 3.08)],
  'phi': [('rnlt_safrane', 3.92), ('vw_passat', 2.5214), ('bmw_320d', 1.8194), ('saab_tid', 0.839), ('audi_a4', -0.27936), ('cit_xantia', -0.5596), ('rnlt_laguna', -1.3995), ('alfa_156', -1.8194), ('peugeot_406', -2.381), ('cit_xsara', -2.661)],
  'matrix': '...'
}

Electre Iv / Is

from EasyMCDM.models.Electre import Electre

data = {
    "A1" : [80, 90,  600, 5.4,  8,  5],
    "A2" : [65, 58,  200, 9.7,  1,  1],
    "A3" : [83, 60,  400, 7.2,  4,  7],
    "A4" : [40, 80, 1000, 7.5,  7, 10],
    "A5" : [52, 72,  600, 2.0,  3,  8],
    "A6" : [94, 96,  700, 3.6,  5,  6],
}
weights = [0.1, 0.2, 0.2, 0.1, 0.2, 0.2]
prefs = ["min", "max", "min", "min", "min", "max"]
vetoes = [45, 29, 550, 6, 4.5, 4.5]
indifference_threshold = 0.6
preference_thresholds = [20, 10, 200, 4, 2, 2] # or None for Electre Iv

e = Electre(data=data, verbose=False)

results = e.solve(weights, prefs, vetoes, indifference_threshold, preference_thresholds)

Output :

{'kernels': ['A4', 'A5']}

Pareto

from EasyMCDM.models.Pareto import Pareto

data = 'tests/data/donnees.csv'
# or
data = {
  "alfa_156": [23817.0, 201.0, 8.0, 39.6, 6.0, 378.0, 31.2],
  "audi_a4": [25771.0, 195.0, 5.7, 35.8, 7.0, 440.0, 33.0],
  "cit_xantia": [25496.0, 195.0, 7.9, 37.0, 2.0, 480.0, 34.0]
}

p = Pareto(data=data, verbose=False)
res = p.solve(indexes=[0,1,6], prefs=["min","max","min"])
print(res)

Output :

{
  'alfa_156': {'Weakly-dominated-by': [], 'Dominated-by': []},
  'audi_a4': {'Weakly-dominated-by': ['alfa_156'], 'Dominated-by': ['alfa_156']}, 
  'cit_xantia': {'Weakly-dominated-by': ['alfa_156', 'vw_passat'], 'Dominated-by': ['alfa_156']},
  'peugeot_406': {'Weakly-dominated-by': ['alfa_156', 'cit_xantia', 'rnlt_laguna', 'vw_passat'], 'Dominated-by': ['alfa_156', 'cit_xantia', 'rnlt_laguna', 'vw_passat']},
  'saab_tid': {'Weakly-dominated-by': ['alfa_156'], 'Dominated-by': ['alfa_156']}, 
  'rnlt_laguna': {'Weakly-dominated-by': ['vw_passat'], 'Dominated-by': ['vw_passat']}, 
  'vw_passat': {'Weakly-dominated-by': [], 'Dominated-by': []},
  'bmw_320d': {'Weakly-dominated-by': [], 'Dominated-by': []},
  'cit_xsara': {'Weakly-dominated-by': [], 'Dominated-by': []},
  'rnlt_safrane': {'Weakly-dominated-by': ['bmw_320d'], 'Dominated-by': ['bmw_320d']}
}

Weighted Sum

from EasyMCDM.models.WeightedSum import WeightedSum

data = 'tests/data/donnees.csv'
# or
data = {
  "alfa_156": [23817.0, 201.0, 8.0, 39.6, 6.0, 378.0, 31.2],
  "audi_a4": [25771.0, 195.0, 5.7, 35.8, 7.0, 440.0, 33.0],
  "cit_xantia": [25496.0, 195.0, 7.9, 37.0, 2.0, 480.0, 34.0]
}

p = WeightedSum(data=data, verbose=False)
res = p.solve(pref_indexes=[0,1,6],prefs=["min","max","min"], weights=[0.001,2,3], target='min')
print(res)

Output :

[(1, 'bmw_320d', -299.04), (2, 'alfa_156', -284.58299999999997), (3, 'rnlt_safrane', -280.84), (4, 'saab_tid', -275.817), (5, 'vw_passat', -265.856), (6, 'audi_a4', -265.229), (7, 'rnlt_laguna', -262.93600000000004), (8, 'cit_xantia', -262.504), (9, 'peugeot_406', -252.551), (10, 'cit_xsara', -244.416)]

Instant-Runoff Multicriteria Optimization (IRMO)

Short description : Eliminate the worst individual for each criteria, until we reach the last one and select the best one.

from EasyMCDM.models.Irmo import Irmo

p = Irmo(data="data/donnees.csv", verbose=False)
res = p.solve(
    indexes=[0,1,4,5], # price -> max_speed -> comfort -> trunk_space
    prefs=["min","max","min","max"]
)
print(res)

Output :

{'best': 'saab_tid'}

List of methods available

Promethee I
Promethee II
Electre Iv
Electre Is
Weighted Sum
Pareto
Instant-Runoff Multicriteria Optimization (IRMO)

Build PyPi package

Build: python setup.py sdist bdist_wheel

Upload: twine upload dist/*

Citation

If you want to cite the tool you can use this:

@misc{EasyMCDM,
  title={EasyMCDM},
  author={Yanis Labrak, Quentin Raymondaud, Philippe Turcotte},
  publisher={GitHub},
  journal={GitHub repository},
  howpublished={\url{https://github.com/qanastek/EasyMCDM}},
  year={2022}
}

Multiple-criteria decision-making (MCDM) with Electre, Promethee, Weighted Sum and Pareto

Related tags

Overview

EasyMCDM - Quick Installation methods

Install with PyPI

Install with GitHub

Setup with Anaconda

Try It

Promethee

Electre Iv / Is

Pareto

Weighted Sum

Instant-Runoff Multicriteria Optimization (IRMO)

List of methods available

Build PyPi package

Citation

Owner

Labrak Yanis

Official PyTorch Implementation of Unsupervised Learning of Scene Flow Estimation Fusing with Local Rigidity

Off-policy continuous control in PyTorch, with RDPG, RTD3 & RSAC

Code for the paper "Relation of the Relations: A New Formalization of the Relation Extraction Problem"

Implementations of orthogonal and semi-orthogonal convolutions in the Fourier domain with applications to adversarial robustness

Active window border replacement for window managers.

The MLOps platform for innovators 🚀

MIMIC Code Repository: Code shared by the research community for the MIMIC-III database

Image Processing, Image Smoothing, Edge Detection and Transforms

learned_optimization: Training and evaluating learned optimizers in JAX

Gender Classification Machine Learning Model using Sk-learn in Python with 97%+ accuracy and deployment

Public Code for NIPS submission SimiGrad: Fine-Grained Adaptive Batching for Large ScaleTraining using Gradient Similarity Measurement

PyTorch ,ONNX and TensorRT implementation of YOLOv4

This is the official repository of the paper Stocastic bandits with groups of similar arms (NeurIPS 2021). It contains the code that was used to compute the figures and experiments of the paper.

A large-scale face dataset for face parsing, recognition, generation and editing.

Modified prey-predator system - Modified prey–predator model describes the rate of change for each species by adding coupling terms.

Ensemble Learning Priors Driven Deep Unfolding for Scalable Snapshot Compressive Imaging [PyTorch]

PyTorch implementation of the ACL, 2021 paper Parameter-efficient Multi-task Fine-tuning for Transformers via Shared Hypernetworks.

Latex code for making neural networks diagrams

Revisiting Weakly Supervised Pre-Training of Visual Perception Models

Detail-Preserving Transformer for Light Field Image Super-Resolution