Prediction of UV/visible absorption maxima of organic compounds in dichloromethane and database generation of organic compounds with red-shifted absorption maxima

Abstract

The current study uses machine learning (ML) to estimate the UV/visible absorption maxima. There are four ML models that are tested. Random Forest model is the best model due to smallest difference between the r-squared values for the training set and test set. Important features are also researched. Python-based tools are used to generate and visualise new chemical compounds, totalling twenty thousand. Predicted UV/visible absorption maxima values are used to screen organic substances. Red-shifted absorption organic molecules are chosen. Analysis of synthetic accessibility scores has indicated that synthesis of large percentage of selected compounds will be easy.