| dc.description.abstract | The spread of the COVID-19 pandemic broughtsignificant changes in society. Emerging technologies like artificial intelligence and machine learning devices improved several industries, especially in academe and higher education institutions. In this study, a model to analyze and predict college students' sentiments from the Flexible Learning Experience portal was built using several supervised machine-learning techniques. Waikato Environment for Knowledge Analysis (WEKA) application was used to apply the Naive Bayes (NB), C4.5, Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN) algorithms. Additionally, a comparative analysis of different machine-learning methods was applied. The experimental results revealed that the C4.5 algorithmobtained the highest accuracy than other algorithms. The effectiveness of each algorithm was evaluated and compared using 10-fold cross-validation (CV), taking into account the major accuracy metrics, instances that were accurately or inaccurately classified, kappa statistics, mean absolute error, and modeling time. Moreover, results show that the C4.5 algorithm outperformed other algorithms by classifying the model with 98.13% accuracy, 0.0132 mean absolute error, and 0.00 seconds of training time. Furthermore, teachers and college administrations were well accustomed to the sentiments and problems of college students and might act as a decision-support mechanism mainly as they deal with the new setting during this time of crisis. | en_US |
