This project aims to propose a module for brightway2 that implements methods of sensitivity analysis proposed in the article of (Wei et al., 2015).
We provide a python module lsa.py
for achieving a local sensitivity analysis.
virtualenv env
source env/bin/activate
pip install -r pip-requires.txt
You should then be able to run LSA example:
python exampleLSA.py
It should write two output files that gives the relative sensitivity coefficients (RSC): exampleLSA_rsca.csv and exampleLSA_rscb.csv
A docker image is available and allow to easily run a notebook with BrightWay2. See the documentation of the jupyter docker image for launching options.
docker pull cmutel/brightway2
docker run -it --rm -p 8888:8888 --volume=$(pwd):/home/jovyan/notebooks cmutel/brightway2
You can also use the enhanced version that I propose, that add the Jupyter notebook extensions with a lot of features for the notebook. For that:
docker build -t bw2 .
docker run -it --rm -p 8888:8888 --volume=$(pwd):/home/jovyan/notebooks bw2
For launching a console and running the samples:
docker run -it --rm --volume=$(pwd):/home/jovyan/notebooks bw2 bash
- Deploy a test instance for easy demonstration and contact Philippe, Pyrenne and Éléonore on 2/06
- example with sample data
- guide for describing a model in brightway2
- Is the factor's selection pertinent after the LSA (threshold=0.1)?
- maybe fix instead a fixed number of factor (for example 50 or 100)
- Some missing uncertainties in section 1.5 (fixme). To see with Chris Mutel
- Which is the best GSA indicator among mu and mu_star? Are the errors relevant on the charts?
- test with higher number of simulations
- Produce a procedure for installation on windows
- import EcoInvent data
- Done: Produce a python module as lsa.py with an easy access to GSA