An Optimal Machine Learning Approach for Interpretation of Emotions and Prediction of School Students’ Attitude towards Online Education as a Post- Effect of Covid-19 in India

The Indian education system was severely impacted during COVID-19 because of the unpreparedness of the entire education arrangements to manage the studies in an online mode. The students particularly senior secondary was deeply shocked when the schools closed down unprecedently and efforts were made by the schools to conduct classes and examinations in an online mode. This study aims to analyze the students’ attitude towards the teaching-learning process considering the availability of digital devices and Internet facilities, before, during, and post covid times. Data from 113 school students have been collected and it comprises twenty independent features and one dependent feature. Exploratory data analysis followed by the application of a Support Vector Machine has been implemented to predict the dependent feature viz. the choice to adopt the online mode of learning. A grid search has been used to tune the threshold and locate the optimal value of the threshold for the ROC Curve. Python scikitlearn's univariate feature selection methods to select the top five features has resulted in the selection of just five independent features that are capable enough to predict the mentioned dependent features with an accuracy of 0.74 to predict the choice to adapt the online mode of learning with Support Vector machine learning classifier. The sentiments and emotions of the students pre-COVID, during-COVID and post-COVID have also been extracted.