A library of ML training of materials' properties

• datasets: python class to download the data from open database + data in json format
• descriptors: python class for different types of descriptors (RDF, ADF, chemical labeling and enviroments)
• ml: python class for the choice of different pipelines of ML methods (KRR, ...)
• test: python class for unit test (to implement)
• results: a collection of results.

This code has the following hierarchy

pyxtal_ml
├── main.py
├── descriptors -> descriptors.py -> (RDF.py, ADF.py, DDF.py, Chem.py, .etc)
├── datasets -> collection.py -> (json files)
├── ml -> methods.py (KRR, KNN, ANN, .etc.)
├── test -> (various scripts to test the accuracy and efficiency of the code)

# git clone https://github.com/qzhu2017/PyXtal_ml.git
# python setup.py install

• SciPy 1.0.1
• NumPy 1.14.3
• Scikit-learn
• Pandas 0.20.3
• Pymatgen