Where Do I Begin? Tuning Support Vector Machines and Boosted Trees

• Activity Number: 660 - Machine Learning: Advances and Applications
• Type: Contributed
• Date/Time: Thursday, August 1, 2019 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Jill Lundell*
• Companies: Utah State University
• Keywords: machine learning; tuning; support vector machines; boosting

Abstract

Support vector machines and boosted trees can be powerful models if they are tuned well. However, hyperparameter tuning can be difficult, particularly for an inexperienced user. This talk introduces an R package that auto-tunes support vector machines and boosted trees. The research behind the package will also be presented to provide insight into the characteristics of hyperparameters. This research can be used to implement a better grid search or develop your own tuning methods.