Using Gradient Methods to Predict Twitter Users' Mental Health with Both COVID-19 Growth Patterns and Tweets

Abstract

Twitter users post tweets to express their feelings, emotions, and behavior. During COVID-19 times, people moved to varied life routines. Such a change in daily life affected people's mental health. We studied the mental health of twitter users during this time through their tweets and compared them with the COVID-19 growth pattern. We also attempted to forecast the depressive tweets and compared them with real data using ARIMA methods. We found our observations of tweets and COVID-19 Epidemic reports of WHO followed a similar pattern. Our forecast findings with ARIMA methods supported the real data.