SENTIMENT ANALYSIS OF PRESIDENTIAL CANDIDATES IN 2019 INDONESIAN GENERAL ELECTION USING MACHINE LEARNING

• Main Author: Adhitama Christanto, Ivan
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/38345
• Institution: Institut Teknologi Bandung

The 2019 Indonesian Presidential General Election – commonly referred to as Pilpres 2019 – is held by KPU (Komisi Pemilihan Umum) to determine Indonesia's president and vice president for 2019-2024. In this election, the registered presidential and vicepresidential candidates were Ir. H. Joko Widodo and Prof. Dr. (HC). K. H. Ma'ruf Amin as number 1 and H. Prabowo Subianto and Sandiaga Salahuddin Uno, M.B.A. as number 2. By knowing the public sentiment towards the two presidential candidates, each team can estimate the possibility of winning the candidate who was promoted in the 2019 Presidential Election and formulate the appropriate campaign strategy. Sentiment analysis is done using Twitter text data relating to the two presidential candidates to find out the sentiments of each candidate. Two machine learning models were built, namely Naïve Bayes Classifier and Support Vector Machine, to classify the Twitter texts into groups of positive, neutral, and negative sentiments. The quality of the two models is compared to determine the best model that will be used as a reference in sentiment analysis. By using the classification results from the best model, sentiment analysis is conducted to determine the acquisition of positive and negative sentiments for each candidate. Also seen is the acquisition of positive and negative sentiments of the two candidates at several different times. At the end of this study, we conclude the winning of one candidate in obtaining positive sentiments compared to other candidates.