This research investigates the improvement of Kernel Density Estimation (KDE) comprehension in a university context via visualization-enhanced teaching. The study tackles KDE misconceptions, the efficacy of visual aids, and the role of previous mathematical and machine learning knowledge. Using a mix of literature review, survey, and experimental study, findings show that visualization techniques notably enhance KDE understanding and application, validating our hypothesis. This work underlines the importance of evidence-based teaching in machine learning education.