Optimizing decision trees for English Teaching Quality Evaluation (ETQE) using Artificial Bee Colony (ABC) optimization

• Published in: Heliyon Vol. 9; no. 8; p. e19274
• Main Author: Cui, Yingying
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.08.2023; Elsevier
• Subjects: algorithms; Artificial intelligence; Language teaching; Meta-heuristic algorithms; prediction; regression analysis; Teaching quality; Teaching quality; Meta-heuristic algorithms; Artificial intelligence; Language teaching
• ISSN: 2405-8440; 2405-8440
• DOI: 10.1016/j.heliyon.2023.e19274

Changes in educational systems and English teaching strategies have increased the need for automatic methods for English Teaching Quality Evaluation (ETQE). A practical model for ETQE applies in different fields, determines the most relevant factors in teaching quality (TQ), and has optimal performance in different conditions. This paper presents a new method based on Artificial Intelligence (AI) and meta-heuristic algorithms to solve the ETQE problem. The proposed method performs the prediction process in two phases: “determination of related indicators” and “quality prediction”. During the first phase, after introducing a set of 24 candidate indicators, an optimal subset of them having maximum correlation with ETQE and minimum redundancy are selected using Artificial Bee Colony (ABC) algorithm. In the second phase of the proposed method, a Classification and Regression Tree (CART) model optimized by ABC are applied to predict ETQ based on the indicators determined in the first phase. In this learning model, split points of decision nodes are determined by ABC in a way that the prediction accuracy would be maximized. The performance of the proposed method has been evaluated in two different teaching environments. The performance of the proposed method has been evaluated in two different teaching environments. The studied teaching environments are face-to-face (FF) and online classes that were held for middle school and university students, respectively. Based on the obtained results, the proposed method can predict the ETQ with an accuracy of more than 98.99% in both tested scenarios, which results in an increase of at least 1.11% compared to the previous methods. The efficiency of the proposed model in both studied scenarios prove the generality of this method to be used in real-world applications.