In extrusion processes, melting strongly influences the development of product end-use properties such as tensile strength. For effective monitoring and control of these properties, it is essential to describe the melting process quantitatively. Obtaining a high-fidelity quantitative description of the melting process, however, remains a challenging problem. This article proposes and employs a hybrid (physical/empirical) modeling approach to model the melting process. The recently developed "pulse perturbation technique" is used to generate the input/output data, while the model structure is determined from physical considerations. Each parameter of the identified model can therefore be associated with a distinct melting mechanism, thus providing valuable quantitative insight into the melting process. Based on the melting data presented in Wetzel et al. [1], the model is used to analyze the effect of extruder operating conditions on the melting of semicrystalline, amorphous, and rubbery polymers. Pulse experiments are conducted on an example reactive extrusion process and a quantitative relationship between the model parameters and end-use properties is developed. Such a quantitative relationship can be used for fast predictions of infrequently measured enduse properties and therefore has potential applications in monitoring and control of product characteristics in extrusion processes.