Polymer Informatics Method for Fast and Accurate Prediction of the Glass Transition Temperature from Chemical Structure' - dataset

The data file attached named "Supporting_data_file.xlsx" includes the full PAEK-Tg data set under the sheet named "Data". All data under "Data" was collected from the literature and from Victrex R&D; corresponding references are provided under the PDF named "Data_references.pdf". Although the SMILES of all homo/co-polymers is provided, a bespoke naming convention was applied for improved/faster visibility of the (sub)structures; the key for this naming convention is found under the sheet named "Key" in the file "Supporting_data_file.xlsx". All count matrices (averaged for copolymers as discussed in the manuscript) are provided under the sheets "X_L_Ar_L", "X_L_Ar" and "X_Ar_L_Ar" corresponding to fragment definitions L-Ar-L, L-Ar and Ar-L-Ar, respectively. All descriptors used for the QSPR-GAP models (i.e. L-Ar-L fragments) are provided in the sheet named "D_L_Ar_L". All descriptors used for the pure QSPR models are provided in the sheet named "D_polymer". The data for figures 2, 3 and 4, is provided in the sheets with prefix "Fig*".