Ballet is a complex art form that consists of precise movements atypical to the human body and requires meticulous technique. Students receive only a limited portion of their instructor's attention in a classroom setting and private coaching is often financially inaccessible, which results in overlooked mistakes. As a result, these momentary errors become habitual and culminate in serious injuries that would be prevented with proper body alignment. This research focused on enhancing the current technology in instructional programs for ballet by training a teacher model with open-source video data of professional dancers. We used a novel Feature Angle Extraction system as a medium to standardize variances among different dancer videos. Our pipeline consisted of the pose estimation system OpenPose to identify dancers’ joint positions, Dynamic Time Warping to align video frames, Random Forests Classifier to feature select characteristics of correct movements, and Decision Tree Classifiers to generate feedback. The outcome of this study was a highly accurate application that computes a percent depiction of movement quality and returns user-friendly feedback upon the input of a single-view video.