The document outlines the development of automated workflows for creating prediction models to assess the bioactivity of small molecules targeting the retinoid x receptor α (rxra), using public-domain data from PubChem. Six machine learning methods were employed to build models, with the best performing model achieving an AUC score of 0.77 and a balanced accuracy (BACC) of 0.70, indicating variations in performance across different external test sets. The study highlights the potential of using publicly available bioactivity data for developing predictive models in chemoinformatics.