COBRA-ML: An Open-Source Web Tool for Multicriteria Decision Making with Automatic Parameter Calibration by Machine Learning

Multicriteria decision-making (MCDM) tools rarely combine objective weighting, automatic parameter calibration, and accessibility for non-programmers in a single open environment. This paper describes COBRA-ML, an open-source web tool developed in Python/Streamlit that integrates the COBRA method with Preference Selection Index (PSI) objective weighting and a Gaussian balancing mechanism whose parameters are automatically calibrated via Random Forest. The bilingual (Portuguese/English) interface requires no local installation and is organized into seven modules covering data input, criteria configuration, pipeline execution, parameter tuning, interactive visualization, generative-AI interpretation, and multi-format export. Three preloaded application domains—renewable energy, military aircraft, and supplier selection—allow immediate use without data preparation. An illustrative case study with renewable energy sources demonstrates the workflow. The tool is registered with INPI and publicly available on GitHub, providing the Operations Research community with a transparent, extensible entry point for COBRA-based decision analysis.