To deliver on the potential outcome-based teaching and learning holds for engineering education, it is important for engineering courses to provide students with different types of deliberate practice opportunities that align to the program’s learning outcomes. Working from these requirements, we increased the design and measurement intentionality of a digital signal processing (DSP) course. To align the course’s learning outcomes more constructively with its assessment measures, we automated the process of classifying DSP questions according to learning outcomes by introducing a model that integrates topic modeling and machine learning. In this work, we explored the effect of pre-processing procedures in terms of stopword selection and word co-occurrence redundancy issue in question classification inferences. In this work, we proposed a customized variant of the Word Network Topic Model, q-WNTM, which is able to use its pre-classified DSP questions to reliably classify new questions according to the course’s learning outcomes.