Exploring the tree algorithms to generate the optimal detection system of students' stress levels

Abstract

The significant changes in the world of education after the coronavirus disease 2019 (COVID-19) pandemic have increased students' anxiety levels. This anxiety can trigger stress which can interfere with students' academic performance. Therefore, this condition is a critical problem that needs to be addressed immediately. However, researchers have not previously conducted much research to detect post-COVID stress levels. Apart from that, the existence of a system capable of carrying out this detection is still lacking. Therefore, this research focuses on building a system for detecting student stress levels. First, an exploration of the tree algorithm was carried out to find the most optimal method for recognizing student stress levels. Then a detection system is built using this optimal method. The research results show that the tree ID3 (Iterative Dichotomiser 3) algorithm achieves the highest accuracy value of 95% compared to other tree algorithms with the scenario of dividing training data into test data of 80%:20%. Moreover, this telegram bot-based detection system works well in recognizing three categories of stress, namely: light-, moderate-, and heavy stress based on black-box testing techniques.