The thesis addresses the question of to what extent we can estimate mathematical anxiety based on visual cues. In addition to programming the camera for detecting skeletal points, the work involves data processing and analysis using various machine learning methods. The design for conducting the experiment and recognizing mathematical anxiety stems from a research project titled "Monitoring and Improving Individual Student Treatments in COVID and Post-COVID Conditions," which aims to assist elementary school students in overcoming anxiety while solving mathematical problems.
During the data processing, we employed various machine learning classifiers and evaluated the performance of individual classifiers for the selected features. In our case, the Support Vector method proved to be the most effective. However, the results of the classifier performance did not prove to be sufficiently accurate to reliably identify mathematical anxiety based solely on visual cues.
|
