Towards Predicting COVID-19 Trends: Feature Engineering on Social Media Responses

Amid the COVID-19 pandemic, the use of social media and virtual networks has been at an all-time high. The motivation of this study is these tweet messages shared by individuals using Twitter could be used to model and predict COVID-19 trends. This study utilized Machine Learning techniques and algorithms such as Linear Regression and Support Vector Machine. To understand the tweet messages, keywords are extracted from the tweets. After analyzing the keywords using the correlation analysis and sentiment analysis, we got the best feature sets to build models to predict the trends of Covid-19 cases and death rate. We build models with analyzed features from COVID-19 daily statistics (baseline) to 47 keywords from tweet messages and compared them with Root Mean Square Error, Mean Absolute Error, and R Square as our model evaluators. We found adding the features selected with correlation and sentiment analysis from the tweets helps to build models greatly and predicted the trends much more precisely with RMSE score of 0.07346836, MAE of 0.0491152 and RSQ 0.374529 (daily cases), and RMSE score of 0.0425504, MAE of 0.03295105 and RSQ of 0.5237014 (daily deaths)