LAUSR

Abstract The field of polymeric materials is one of the most complex that exists. These materials have very high molecular weights and also a molecular weight distribution, which give them singular properties. The demand for new materials that suit specific applications is increasing. However, the development time of new materials from new molecular structures can take 10–20 years. For this reason, both the knowledge about the structure-property relationship and the creation of reliable databases are increasingly crucial, as they serve to generate predictive models with the aim of reducing development times. This challenge is attempted by polymer informatics. In the present work, we show results of in silico experimentation, using in-house databases, with the aim of generating Quantitative Structure-Property Relationship (QSPR) models to predict a property derived from the tensile test Tensile Modulus. Two models were reported, after several development stages: feature selection, QSPR modelling training and validation. Additionally, complete physicochemical discussions and interpretations were presented. The QSPR model could be used as a “virtual testing” that provides a property profile estimation for a new molecule before synthesis. Consequently, our model is expected to contribute in the design stage of new polymeric materials, drastically reducing costs and development times.