MLcps: machine learning cumulative performance score for classification problems.

Area of Science:

• Computer Science
• Data Science
• Machine Learning

Background:

• Assessing machine learning (ML) model performance necessitates multiple evaluation metrics for a comprehensive understanding.
• Individual metric comparison for model selection is time-consuming and prone to user bias.

Purpose of the Study:

• Introduce the Machine Learning Cumulative Performance Score (MLcps) as a novel, unified evaluation metric for classification models.
• Provide a holistic approach to assessing ML model performance, integrating various metrics into a single score.

Main Methods:

• Developed MLcps, a novel metric that consolidates multiple precomputed evaluation metrics.
• Tested MLcps on four publicly available datasets for classification problems.

Main Results:

• MLcps provides a unified score for comprehensive model assessment, highlighting strengths and weaknesses.
• Demonstrated MLcps's ability to offer a holistic evaluation of model robustness and overall performance across datasets.

Conclusions:

• MLcps streamlines the model evaluation process by replacing the need to compare individual metrics.
• Researchers and practitioners can efficiently assess ML models using a single MLcps value, saving time and effort.
• MLcps is available as an open-source Python package for broader accessibility.