Activity based analysis and prediction strategy for the class room performance improvement

Abstract

Class room assessment is one of the key issues where in individual performance gives a great impact in the overall performance of the class. In general, various activities will help in identifying the student’s ability and weak strengths in understanding the subject as well. Though various online learning platforms are present, an overwhelmed usage is occurring due to the pandemic in this period. Nevertheless, very few online resources are giving their utmost use in efficient way. At the same time class room activities are also need to shape up accordingly that supports to the virtual platforms. Few activities like chitchat, polling, and turn around were performed through virtual platform over a class group, to address the complex engineering problems that were identified in Embedded Systems Design, an undergraduate subject. Further a common strategy was found to apply while dealing with that subject. In this process, collected the outcomes of the complete learning activities conducted during the course and using few data-mining techniques the classroom performance has been analysed. For initial analysis, k means clustering was used in which the parameters chosen are like individual marks, total class strength. The outcome of this implementation resulted in identification of clusters in the name of weak, average and best groups along with an accuracy of 46.51%. The analyses were further carried out using other measuring parameters like spontaneity, delays in answering, best answer, etc., have been chosen during the activities and Naïve Bayes Classifier was chosen for predicting the performance improvement and was observed as 70%. Finally, based on the experimental results, classification and regression supplemented by logistic was applied for further evaluation of learning performances and a remarkable accuracy progress of 96.96% is investigated in this study.