Machine learning models trained on the client's polymer data guided the design and optimization of new polymers, with later measurements confirming the predictions.
Challenge
The project used client-provided polymer data and measured properties, including tensile strength, to guide the design of new polymers for human health applications.
What We Did
-
1
Built property-prediction models
Used the client's polymer data to train support vector regression (SVR) and random forest regression (RFR) models for the measured properties.
-
2
Validated model performance
Evaluated the models using data scrambling and 5-fold cross-validation.
-
3
Designed and optimized polymers
Used predictions for monomers and the property models to guide the design and optimization of novel polymers.
-
4
Worked in a rapid feedback cycle
Completed the project in three weeks with weekly updates and reports. The client later synthesized the polymers and measured polymer properties, confirming the predictions.