Abstract
The dataset "Dye-Compounds.csv" is a comprehensive molecular collection of 8,815 dye compounds, each annotated with detailed physicochemical and structural descriptors. It includes 248 features such as atom counts (C, H, N, O, S, halogens, metals), RDKit-computed molecular descriptors (molecular weight, logP, ring counts, topological indices, valence electrons, aromatic heterocycles), and fragment-based structural fingerprints. The target variable of interest is Epsilon (molar extinction coefficient), which quantifies light absorption capacity of dyes, making it a critical property for optical and materials applications. Additionally, derived meta-features such as conformer counts, aromatic atom counts, and longest carbon chain length provide rich structural diversity. The dataset is highly multidimensional, including numeric and categorical descriptors, and is appropriate for quantitative structure–property relationship (QSPR) modeling, cheminformatics, and machine learning-based predictive dye design.