This repository belongs to the the publication "Interpretable machine learning for predicting membrane performance of organic solvent nanofiltration membranes using an open-access database" published in the Journalof Membrane Science (DOI: 10.1016/j.memsci.2026.126088).
It uses a dataset derived from The Open Membrane Database (https://openmembranedatabase.org/). The access date was April 8th 2025. The original dataset is in the repository as OMD_SRNF_2025-04-08.csv file.
The main researchers involved in the study are Sarah Glass (ORCID: 0000-0002-0625-0057), Petra Merten (no ORCID available), Torsten Brinkmann (ORCID: 0000-0002-5114-6259) and Volkan Filiz (ORCID: 0000-0003-0239-8641).
The repository contains 5 Python files:
- Dataset_SRNF.py (Statistical analysis of the original dataset and the subsets.)
- data_preprocessing.py (Preprocessing of the original dataset before all modelling activities.)
- MWCO_Model.py (Trained models for the MWCO subset.)
- Perm_Model.py (Trained models for the solvent permeance subset.)
- plots.py (All plots not directly generated by the models.)
The analysis was performed using Python Version=3.13.5
Required packages include:
- pandas (Version=2.3.2)
- numpy (Version=2.2.6)
- scikit-learn (Version=1.7.1)
- matplotlib (Version=3.10.5)
- seaborn (Version=0.13.2)
- shap (Version=0.48.0)
See Contributing.md
See Citation.cff
This project is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
See the LICENSE file for details.