Characterization data conversion for a product prediction model

Abstract

In at least one example, a method can include determining first parametric model parameters for characterization data of a first type for a first plurality of polymers, determining second parametric model parameters for characterization data of a second type for the first plurality of polymers, correlating the first parametric model parameters to the second parametric model parameters, converting, based on the correlation, characterization data of the first type for a second plurality of polymers to characterization data of the second type for the second plurality of polymers, training a product prediction model with the characterization data of the second type for the second plurality of polymers, wherein the product prediction model was previously trained with characterization data of the second type for a third plurality of polymers, and performing, by the product prediction model, a prediction of a property of a polymer that is not one of the first, second, or third pluralities of polymers.

Images