Code used in Methodological Framework for Materials Discovery using Machine Learning paper.

The preprocessed data for the first case study (vdW vs nvdW modeling) is located in data\vdW_nvdW\preprocessed_materials_project_data.csv. Similarly, the preprocessed data for the second case study (wide bandgap vs non-wide bandgap) is located in data\UV_nUV\preprocessed_vdW_data.csv.

To repeat the first case study experiment, run the python scripts in scripts\vdW_nvdW_classification. To repeat the second case study experiment, run the python scripts in scripts\UV_classification.

The software used to run the self-organizing map (SOM) algorithm in this project is licensed and hence cannot be posted in public. Instead, we suggest several alternative SOM packages that can be used to run SOM modeling (although they are not tested for this project):

• https://github.com/sevamoo/SOMPY (for Python)
• http://www.cis.hut.fi/somtoolbox/ (for Matlab)