E-LEARNING TWITTER SENTIMENT ANALYSIS USING SUPPORT VECTOR MACHINE: A COVID-19 CASE STUDY

Abstract

It has been more than a year since the Covid-19 pandemic struck. It is evident that adoption of online teaching has changed the landscape of the education sector. However, the efficacy of online teaching is the main concern because it depends on the study of the present time as the method is evolving. Another problem is platform such as Sistem Pengurusan Aduan Awam (SISPAA) that exist to accept complaints and suggestions from the people is found to be very inefficient since not many people are familiar with it. Furthermore, it is quite complex to interpret the sentences and analysing the sentiments due to the granularities of words and phrases. As a solution, Sentiment Analysis can be conducted to analyse the sentiments of the people on Twitter platform. These sentiments will reflect the view and the efficacy of the new method by providing statistical analysis on the issue. Lexicon and rule-based tool, VADER is used to label the sentiments that later will be visualized, and further steps are taken to test the accuracy of Support Vector Machine (SVM) model against Linear Regression (LR) model. The analysis shows majority of the twitter users have a positive review for E-learning and the SVM is the better classifier because it has a higher accuracy percentage compared to LR. This project follows Data Science Life Cycle methodology and confusion matrix is used to visualize the accuracy testing for both models. The significance of this project is to provide a statistical analysis on the new approach based on Twitter data. Plus, the findings of this project can be used to improve the implementation of current online teaching method. For future research, Sentiment Analysis can be improved by tuning different parameter property and using another various technique with different machine learning models.