Tg_Benchmarking

Code and data for the paper Benchmarking Machine Learning Models for Polymer Informatics: An Example of Glass Transition Temperature.

Code

Code to compare the true Tg and predicted Tg from differnet machine learning models.

Data

• All the polymer data can be found in the PolyInfo database https://polymer.nims.go.jp/en/ (National Institute for Materials Science (NIMS) holds the copyright of this database system).
• MD simulated 566 polymers in Dataset_2 of the paper are listed in csv file.
• MD simulated 100 polymers in Dataset_1 of the paper are listed in excel file.

Model

Keras and scikit-learn models are saved into pickle files.

Packages

Required packages and versions are listed in the requirements.txt.