Dataset and Python-based Machine Learning tools for data-driven modeling of water adsorption isotherms in cocoa beans

This dataset provides experimental water adsorption isotherms for dried and roasted cocoa beans (Theobroma cacao L.) and Python-based machine learning tools for data-driven modeling. Isotherms were measured using the Dynamic Dewpoint Isotherm (DDI) method across a water activity range of 0.1-0.85 and temperatures of 25, 30, and 40 °C. The dataset includes raw and organized Excel files containing equilibrium moisture content, water activity, experimental conditions, and replicates, enabling reproducible modeling. Fully documented Python scripts implement Support Vector Regression, Random Forest, and Artificial Neural Networks for predicting moisture content as a function of water activity, temperature, and cocoa type, supporting multivariate analyses. This database facilitates prediction of cocoa beans’ moisture content, assesment of storage, and optimization of post-harvest storage strategies for researchers, producers, and industry stakeholders.