What question did this study set out to answer?

The aim is to present an open-source package for implementing multi-view stacking classifiers in supervised learning tasks.

June 3, 2026Journal of Open Research Software0 citationsOpen Access

multiviewstacking: A Python Package for Training Multi-View Stacking Classifiers

EGEnrique Garcia-Ceja

Key Points

The aim is to present an open-source package for implementing multi-view stacking classifiers in supervised learning tasks.
Developed a Python package supporting multiple models including scikit-learn.
Incorporated data validation and quality-control tests for robustness.
Provided an example dataset for practical experimentation.
Facilitates the development of heterogeneous multi-view classifiers.
Supports applications in diverse fields like activity recognition and cybersecurity.
Enables researchers to enhance ensemble learning techniques effectively.

Abstract

The multiviewstacking Python package provides an open-source implementation of the Multi-View Stacking (MVS) algorithm for supervised classification tasks. MVS extends traditional stacked generalization by training separate first-level learners on multiple complementary views of the same data and combining their predictions through a meta-learner. The package is designed with flexibility and interoperability in mind, supporting any combination of scikit-learn or custom models. It includes built-in data validation, quality-control tests, and an example multi-view dataset for quick experimentation. multiviewstacking enables researchers to easily build and evaluate heterogeneous multi-view classifiers, facilitating data fusion and ensemble learning applications across diverse domains, such as activity recognition, cybersecurity, and biomedical data analysis. The code and examples of the package can be found at: https://github.com/enriquegit/multiviewstacking.

Ask AI

Helpful

Bookmark

View Full Paper