SPlit: An Optimal Method for Data Splitting

2020/12/31 by V. Roshan Joseph, Akhil Vakayil · 198 citations
Computer Science · Mathematics · #Categorical variable #Face and Expression Recognition #Imbalanced Data Classification Techniques #Machine Learning and Data Classification #Point (geometry) #Regression #Regression analysis #Training set #cs.LG #k-nearest neighbors algorithm #stat.ML

paper · pdf · doi:10.1080/00401706.2021.1921037

published in Technometrics 64(2), 166-176 (Taylor & Francis)

openalex created_date 2021/01/05 · arxiv created 2021/03/19 · openalex publication_date 2021/04/28 · arxiv updated 2021/05/10 · openalex updated_date 2026/08/06

Abstract

In this article, we propose an optimal method referred to as SPlit for splitting a dataset into training and testing sets. SPlit is based on the method of support points (SP), which was initially developed for finding the optimal representative points of a continuous distribution. We adapt SP for subsampling from a dataset using a sequential nearest neighbor algorithm. We also extend SP to deal with categorical variables so that SPlit can be applied to both regression and classification problems. The implementation of SPlit on real datasets shows substantial improvement in the worst-case testing performance for several modeling methods compared to the commonly used random splitting procedure.

Citations

• Computer Aided Design of Experiments
• Modeling of strength of high-performance concrete using artificial neural networks

Cited by

• Supervised Stratified Subsampling for Predictive Analytics
• Ten simple rules for reporting machine learning methods implementation and evaluation on biomedical data
• Sliced Space-filling Design with Mixtures
• Slicing the past to predict the future: Recasting data slicing as curatorial work in ML development and evaluation
• Harnessing Near-Infrared Spectroscopy and Machine Learning for Traceable Classification of Hanwoo and Holstein Beef
• Personalized Control for Lower Limb Prosthesis Using Kolmogorov-Arnold Networks
• Predictive Subdata Selection for Computer Models

Related

• The Power of Unbiased Recursive Partitioning: A Unifying View of CTree, MOB, and GUIDE
• Training and Testing with Multiple Splits: A Central Limit Theorem for Split-Sample Estimators
• To Split or Not to Split: The Impact of Disparate Treatment in Classification
• Penalized Split Criteria for Interpretable Trees
• Similarity Based Stratified Splitting: an approach to train better classifiers