ClassifyR: an R package for performance assessment of classification with applications to transcriptomics

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• Existing R packages for classification lack a unified framework for custom functions and performance evaluation.
• There is a need for a standardized approach to reproducible classifier benchmarking.

Purpose of the Study:

• To design and implement a generic, object-oriented R package for classification.
• To facilitate the linking of custom classification functions with performance measures.
• To enable rapid and reproducible benchmarking of alternative classifiers.

Main Methods:

• Developed a generic classification framework as an R package using an object-oriented style.
• Emphasized parallel processing, reproducibility, and extensibility in the package design.
• Integrated a comprehensive set of performance measures for post-processing.

Main Results:

• The ClassifyR package offers a flexible framework for integrating diverse classification algorithms.
• The package supports parallel processing for enhanced computational efficiency.
• Reproducible benchmarking is achieved through standardized linking of functions and measures.

Conclusions:

• ClassifyR addresses the need for a unified and extensible classification framework in R.
• The package promotes reproducible research by standardizing classifier evaluation.
• It empowers researchers to efficiently benchmark and select optimal classifiers for their data.