The models were created utilizing ensemble algorithms and incorporate fingerprints as well as CDK descriptors.

Users are required to input the SMILES of the hydrogen bond acceptor (HBA) and the hydrogen bond donor (HBD), along with the molar fraction of their combination. The solubility of CO₂ is then predicted based on temperature and pressure. Viscosity and density predictions are made solely based on temperature at atmospheric pressure.

The QSAR models are created to predict the minimum inhibitory concentration (MIC) of ionic liquids against three bacteria: Staphylococcus aureus, Escherichia coli, and Pseudomonas aeruginosa. These models utilize a random forest algorithm and are built upon a dataset comprising over 800 experimental data points.

The models are developed using ensemble algorithms and utilize fingerprints as descriptors. These models are constructed using large datasets consisting of: 4708 data points for electrical conductivity, 18324 data points for dynamic viscosity, 31026 data points for density, 6788 data points for surface tension, and 5702 data points for sound velocity.

These models are able to predict melting, thermal decomposition and glass transition temperatures of ionic liquids. The models utilize melting temperature data for 3076 salts, thermal decomposition temperature data for 2780 salts, and glass transition temperature data for 794 salts. These models were constructed using the CatBoost algorithm with Morgan descriptors as inputs.