A High-Performance and Interpretable pKa Prediction Framework Integrating Count-Based Fingerprints and Ensemble Learning

This dataset supports the research titled “An Interpretable pKa Prediction Model Based on Count-Based Morgan Fingerprint (C-MF) and Ensemble Learning.” It aims to provide high-quality annotated data for machine learning predictions of pKa values for organic compounds, validate the model's predictive accuracy and generalization capability, and clarify the model's applicability boundaries. The core objective is to develop a lightweight, highly accessible pKa prediction tool (requiring no quantum chemical calculations, with features rapidly generated solely from SMILES) for large-scale pKa screening in scenarios such as environmental pollutant risk assessment and drug molecule design.The dataset is organized into two categories: “Model Construction Foundation Data” and “Generalization Capability Validation Data,” comprising four files.