Twitter Data Analysis Using Machine Learning To Evaluate Community Compliance in Preventing the Spread of Covid-19

Abstract

The Covid-19 pandemic that has hit the world, including Indonesia, has forced the government to take policies to prevent the spread of this deadly virus. One of the efforts is socialization to gain people's awareness to keep their distance and stay home. One of the media that can be used for this purpose is Twitter. However, even socialization efforts have been carried out, the spread of Covid-19 cases still has not yet decreased. Many aspects have to be evaluated to fix the situation. One of them is to evaluate the level of community compliance compared to the spread of Covid-19. This study aims to determine the level of community compliance to stay at home and its correlation with the number of positive cases of Covid-19 in Indonesia. This study takes the data from the tweets of the Indonesian people. The algorithms used are logistic regression and random forest combined with the ensemble algorithm, namely bagging. Twitter data taken are those that contain the word covid19, tetapdirumah, stayathome, mudik, psbb; location in Indonesia within a period of March 3rd to July 7th 2020. The data obtained from 705 tweets shows that non-compliance community has increased starting mid-June 2020 which is in line with the increasing data trend on the Covid-19 cases in Indonesia.