Sentiment Analysis on MySejahtera Application during COVID-19 Pandemic

MySejahtera is an application developed by the Malaysian government to assist and monitor the Covid-19 flare-up that happened in Malaysia by allowing users to assess their health risk in the event of an outbreak. In MySejahtera, the public obtains significant information about the affected areas and other health-related conditions. The use of MySejahtera became mandatory during the pandemic to reduce the Covid-19 spread. This paper reports the public sentiment towards MySejahtera using the supervised Machine Learning (ML) method. The total number of feedback scraped from Twitter, Google Play Store, and App Store was 33 376, and the modelling was trained using 20 647 feedbacks based on the timeline during the launch of the application; 26^th March 2020 until 25^th November 2021. The corpus was used to determine the feedback label, either positive or negative. The tested ML algorithms are Support Vector Machine (SVM) and Naïve Bayes (NB) through evaluation metrics that are accuracy, precision, recall/sensitivity, and specificity. The result of modelling shows that the SVM classifier with 90:10 percentage split, using Vader extraction technique, has the highest accuracy of 89.93% and recall/sensitivity of 90.55%. The results of this sentiment analysis are visualized for better understanding. It is preferable to utilize a Malay Language corpus and to have more records from Twitter for future work.