Implementasi Scrapping Data Untuk Sentiment Analysis Pengguna Dompet Digital dengan Menggunakan Algoritma Machine Learning

Abstract

Analysis of social media is an effective tool for understanding people's attitudes, preferences and opinions. For a company, social media analysis can help companies to make decisions about needs, attitudes, opinions or trends about potential customers or potential customers. Hootsuite's Wearesocial research results show that out of 150 million Indonesians, 88% are actively using social media. As a result, a lot of information that was once difficult to obtain is now very easy to obtain. With so much data, data collection can be made easier by using an automated data collection system. The use of Machine Learning Algorithm as a method is used to find out how far the level of search from social media data collected is to get accurate results related to changes in behavior, lifestyle, and community activities in dealing with this pandemic. Scraping data is done using Twitter social media data with various hashtags that support opinion mining contained in the community. The results obtained are the analysis of gopay users having a positive sentiment level of 79,6%, negative sentiment of 20,4% from total of 250 data. Meanwhile, Linkaja users have a positive sentiment level of 62,1%, 37,9% negative comments from 250 data comments taken. The test results also carried out a process of calculating the level of accuracy with recall and precision, namely 87% for Gopay and 89% for LinkAja.