Sentiment Analysis Covid-19 Spread Tracing on Google Play Store Application

Abstract

Sentiment analysis of users of tracing the spread of Covid-19 using Google Playstore application review in Southeast Asia, especially the “Peduli Lindungi” application in Indonesia, the “Trace Together” application in Singapore, “My Sejahtera” in Malaysia. The dataset used is a total of 6000 reviews from each application of 2000 reviews during the period June to December 2021. Sentiment analysis classification uses random forest algorithm and logistic regression resulted negative sentiment dominant. Sentiment positive vs negative for the “Peduli Lindungi” was 29% vs 71%, the “Trace Together” was 25% vs 75%, and “My Sejahtera” was 32% vs 68%. Classification performance checked by confusion matrix, logistic regression and random forest resulted in almost the same accuracy, but logistic regression was better with details of accuracy 87%, 84%, and 85%, precision 89%, 85%, 85%, F1 score and recall 86%, 84%, 85% respectively for the “Peduli Lindungi”, the “Trace Together”, and “My Sejahtera” applications, respectively.