The Materials Simulation Toolkit for Machine Learning (MAST-ML): Automating Development and Evaluation of Machine Learning Models for Materials Property Prediction
Category
Published on
Abstract
This tutorial contains an introduction to the use of the Materials Simulation Toolkit for Machine Learning (MAST-ML), a python package designed to broaden and accelerate the use of machine learning and data science methods for materials property prediction. Through hands-on activities, we will use MAST-ML to:
- import materials datasets from online databases and clean and examine our input data,
- conduct feature engineering analysis, including generation, preprocessing, and selection of features,
- construct, evaluate and compare the performance of different model types and data splitting techniques, and
- conduct a preliminary assessment of model error analysis and uncertainty quantification (UQ).
MAST-ML resources:
- MAST-ML on nanoHUB: https://nanohub.org/tools/mastmltutorial
- MAST-ML code: https://github.com/uw-cmg/MAST-ML
- Publication: https://doi.org/10.1016/j.commatsci.2020.109544
- MAST-ML tutorials: https://github.com/uw-cmg/MAST-ML/tree/master/examples
Sponsored by
Cite this work
Researchers should cite this work as follows: