This repository provides four Jupyter notebooks forming a structured learning module on polymer informatics and machine-learning workflows for polymer science. The material progresses from core concepts to predictive modeling, representation analysis, and generative design strategies.
01_polymer_ml_foundation.ipynb
Foundational introduction to polymer informatics, molecular representation, and feature engineering.
Fig 1. Distribution of polymer glass-transition tempeature (Tg) showing dataset coverage and density.
Fig 2. Pearson correlation matrix of molecular descriptors higlighting redundancy and feature dependence.
Fig 3. Distribution of molecular weights of repeating units.
Work in Progess
02_polymer_prediction.ipynb
Polymer property prediction using supervised learning models and evaluation protocols.
03_polymer_feature_space_exploration.ipynb Construction of virtual polymers from small-molecule datasets and polymerization rules, followed by feature-space analysis using dimensionality reduction (t-SNE), clustering, and explainable ML (SHAP).
04_polymer_inverse_design_generation.ipynb
Generative and inverse-design approaches for creating hypothetical polymers using reinforcement-learning-driven optimization.
Last update: 14/02/26