| Summary: | Students' confidence level when responding to questions is significant in assessing their academic performance. Confidence can be defined as a psychological state by a sense of assurance and positive feelings towards one's actions or beliefs. Confidence-based assessment (CBA) evaluates a student's confidence level or expectation concerning their response to identify their actual knowledge. Nevertheless, it should be noted that including imprecise self-learning insights in CBA might lead to a lack of dependability and precision in its measurements. Consequently, this can have a negative impact on the overall effectiveness of CBA. While machine learning (ML) has demonstrated the ability to achieve high accuracy in classification and prediction tasks, it is important to acknowledge the presence of an overfitting issue associated with these methods. Thus, this research aims to describe the steps involved in the conceptual model development to enhance CBA accuracy. The indicators that play significant roles in CBA for developing the conceptual model are multiple choice question correctness answers and selection of confidence level indicators (full or partial). This research involved three phases in developing the proposed conceptual model: 1) problem assessment and research study, 2) indicators of CBA, and 3) ensemble learning approaches to enhance CBA accuracy. Therefore, this proposed conceptual model overcomes the overfitting problems commonly encountered in ML applications. As a result, student's performance can be displayed as the educator can evaluate their teaching methods, and students can recognize the areas that they need to improve. © 2023 IEEE.
|
|---|