ABSTRACT
Online learning is a paradigm shift from traditional offline education;recently there has been a remarkable surge in e-learning platforms due to Covid 19 outbreaks. There is a significant difference in students' performance on both platforms. The primary focus of this study is to investigate how the students perform in both learning methods. Moreover, five ensemble-learning approaches are compared to predict student performance in online and offline education platforms. Ensemble learning is a prominent machine learning meta-approach that integrates predictions from several models to improve prediction. Students' performance data for both offline and online platforms were extracted from a private university's student database. Five ensemble-learning methods were applied to both datasets for predictive analysis. According to the findings of this study, students do better on online platforms than in traditional education systems. Furthermore, XGBoost, Gradient Boost, and Stacking KNN fared better for online data, whereas stacking neural networks and stacking random forest performed better for offline data. The findings of this study will assist educational instructors to concentrate more on students' performance based on their particular learning system. © 2022 ACM.
ABSTRACT
Online based education was enforced during the COVID-19 pandemic to minimize interruption in the pedagogy of teaching. Impact of transition from face-to-face to online based education on students’ learning needs to be studied. This research aims to compare the continuous assessment performance of students in both gmeet-based online and traditional offline classroom methods. Data required for this study was obtained from an academic institution (TCE-Madurai). Three core courses namely Database Management System (IV semester), Web Programming (V semester), and Theory of Computation (III semester) in the Department of Information Technology for both online (2018-2022 batch and 2019-2023 batch) and offline (2017-2021 batch) approach has been compared while considering the same cohort. Each course is designed with Course Outcomes (COs). We did Extrapolative analysis, Descriptive analysis, Correlation analysis, Regression analysis, ANOVA, and MANOVA analysis to find the relationship between internal assessments and terminal examination. The dataset has been collected from 337 students’ from the 2018-22 batch and 134 students from 2017-21 batch in the B.Tech program. Internal assessment includes three continuous assessment tests (CATs) and three assignments marks mapped with respect to corresponding COs. Terminal exam marks were also considered in the study. The dataset contains various features like scores of individual students in both internal assessments and terminal exams. Regression analysis helped to derive the relationship between final exam score which includes a specific course, a specific student, and a specific mode of delivery-Online / Offline. It explained the relationship between final exam score and internal assessment marks. It is observed that the students in online mode had significantly higher scores for the majority of the Course Outcomes, mainly at the Knowledge level. It is difficult to measure certain areas like students’ skills and attitude level such as students’ peer Engagement, Human Interaction, Communication/Preparation, Critical thinking, and Team skills in online mode. © 2022, Rajarambapu Institute Of Technology. All rights reserved.