Topic-Based Classification of MAUDE Adverse Problem Reports through Multi-Application Machine Learning

A key use for warranty claims and repair data is to give input to stakeholders involved in the product life cycle such as design engineering, quality engineering, and management. Text categorization is extensively employed for organizing digital materials. This study aims to make a significant contribution to the field of warranty data analysis by presenting a novel methodology for classifying warranty data reports by stakeholder. The praxis was divided into two parts: dimensionality reduction for feature extraction and predictive model construction for classification. For dimensionality reduction, several approaches such as PCA, T-SNE, LDA, and NMF were used, and the results were assessed using clustering and visualization tools. The T-SNE model with DBSCAN clustering was determined as the best method for labeling the data. Seven classification models were evaluated in the second phase using the accuracy, recall, precision, and F1-score metrics; the Support Vector Machine (SVM) outperformed the others. The XG Boost model’s performance scores were comparable to those of the SVM model and outperformed the SVM model in terms of time performance. Overall, the study demonstrates that the proposed model has excellent classification efficacy as determined by accuracy, precision, recall, and F1-score metrics.

Author

• Bowden Jenkins, Nicole

Language

• en

Keyword

• Classification
• Warranty Analysis

Date created

• 2023

Type of Work

• Dissertation

License

• All rights reserved

Rights statement

• In Copyright

GW Unit

• Engineering Management

Degree

• D.Engr.

Advisor

• Etemadi, Amir

Committee Member(s)

• Blackford, Joseph P.
• Ali, Daniyal

Persistent URL

•  https://scholarspace.library.gwu.edu/etd/cv43nx65m