College
College of Engineering and Polymer Science
Date of Last Revision
2025-04-26 12:11:25
Major
Chemical Engineering
Honors Course
CHEE 497-001
Number of Credits
3
Degree Name
Bachelor of Science
Date of Expected Graduation
Spring 2025
Abstract
The focus for this study is prediction of glass transition temperatures of polymers using machine learning models. The preprocessing of the data included the generation of descriptors, the scaling of the data, the principal component analysis for dimensionality reduction, and the clustering of the data. Three methods of property predictions were utilized including Linear Regression, Random Forest, and a Feedforward Neural Network. The dataset consisted of over 7000 polymers each with their respective glass transition temperatures. This study found that the Random Forest model returned the best results followed by the Feedforward Neural Network then the Linear Regression model. Each model performed reasonably well in the prediction of glass transition temperatures. The models described in this system can be utilized for other properties for both polymers and other systems due to the generalized structure of the model.
Research Sponsor
Fardin Khabaz
First Reader
Bi-min Zhang Newby
Second Reader
Linxiao Chen
Honors Faculty Advisor
Bi-min Zhang Newby
Proprietary and/or Confidential Information
No
Recommended Citation
Wolfe, Nicholas, "Prediction of Glass Transition Temperature of Polymers using Structure-Based Models" (2025). Williams Honors College, Honors Research Projects. 1975.
https://ideaexchange.uakron.edu/honors_research_projects/1975
Included in
Other Chemical Engineering Commons, Other Computer Engineering Commons, Other Materials Science and Engineering Commons, Polymer and Organic Materials Commons, Polymer Science Commons