Description
Abstract: We explore the role of data complexity focusing on how structural characteristics of tabular datasets influence both machine learning algorithm performance and feature selection stability. Using a set of complexity metrics and multiple algorithms, this work reveals that certain data traits, such as dimensionality, overlap, and network structure, can significantly affect predictive stability and effectiveness of the classification algorithms. Knowing the complexity of the data can offer practical insights for selecting appropriate models and feature selection strategies based on dataset properties, contributing to more robust and interpretable defect prediction systems.
Bio: Daniel Rodriguez is currently an associate professor at the Computer Science Department of the University of Alcala, Madrid, Spain. Previously, he was a lecturer at the University of Reading, UK. Daniel earned his degree in Computer Science at the University of the Basque Country (EHU) and PhD degree at the University of Reading, UK. His research interests include data mining and software engineering in general and the application of data and optimisation techniques to Software Engineering in particular.