The advancement of technology has transformed how people interact and express opinions on social media platforms. This research examines Twitter conversations regarding Indonesia's government-initiated Housing Savings Program (TAPERA) through sentiment analysis. The study employed Naïve Bayes classification methodology, with data acquisition conducted via Google Colab platform utilizing the tweet-harvest library. The collection process yielded 1,800 tweets matching predetermined search parameters. Data underwent rigorous preprocessing, including text cleaning and manual sentiment annotation to establish reliable training datasets. Examination of 720 test tweets revealed 473 (65.69%) expressed negative sentiment while 247 (34.31%) conveyed positive sentiment toward the program. The implemented Naïve Bayes model achieved 84.17% accuracy, with negative class precision at 88.71% and recall at 88.60%, while positive class precision reached 78.54% with 76.08% recall. Results indicate the Naïve Bayes approach effectively categorizes public sentiment regarding the TAPERA program, offering valuable feedback for stakeholders responsible for program assessment and enhancement.
Dewy et al. (Fri,) studied this question.