Open Polymer Prediction 2025

Predicting polymer performance from chemical structure using machine learning
Accelerating sustainable materials discovery with open data

---

Project Overview

Polymers are the backbone of innovation in medicine, electronics, and sustainability. Our goal is to predict a polymer's real-world performance directly from its chemical structure.

This project supports the Open Polymer Prediction 2025 initiative by leveraging machine learning on a large-scale, open-source dataset — ten times larger than any existing resource.

We predict five key physical properties from SMILES strings:

• Density
• Thermal conductivity (Tc)
• Glass transition temperature (Tg)
• Radius of gyration (Rg)
• Fractional free volume (FFV)

---

📁 Project Structure

.
├── data/                   # Input datasets (SMILES, labels)
├── src/
│   ├── preprocessing.py    # Molecular parsing and feature extraction
│   ├── models/             # ML/DL model definitions
│   ├── train.py            # Training pipeline
│   ├── evaluate.py         # Evaluation scripts
├── notebooks/              # Jupyter notebooks for EDA and prototyping
├── requirements.txt        # Python dependencies
└── README.md               # Project documentation