There Has to Be an Easier Way: a Simple Alternative for Parameter Tuning of Supervised Learning Methods

• Activity Number: 172 - Machine Learning and Algorithms
• Type: Contributed
• Date/Time: Monday, July 31, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Computing
• Author(s): Jill Lundell*
• Keywords: supervised learning ; support vector machines ; statistical learning ; R ; boosted trees ; optimization

Abstract

Several R packages can tune supervised learning methods, but some packages are so comprehensive they are difficult to use. Others are easier to use, but will only tune one or two methods. This paper presents an alternative R package that uses an optimizer to remove much of the frustration with parameter tuning for gradient boosting machines, support vector machines, and adaboost.